diff --git a/results/tmol/1.0.0.txt b/results/tmol/1.0.0.txt new file mode 100644 index 0000000..0e8a788 --- /dev/null +++ b/results/tmol/1.0.0.txt @@ -0,0 +1,77 @@ +--- +tool: tmol +version: 1.0.0 +image: ghcr.io/brineylab/toolshed/tmol@sha256:97cdbaa7c71dfd694f15137702b15f5fcd2b3af5ef6df5edf49aaccb880d04b0 +digest: sha256:97cdbaa7c71dfd694f15137702b15f5fcd2b3af5ef6df5edf49aaccb880d04b0 +runner: sbatch --partition=l40s --gres=gpu:1 --time=00:30:00 +host: bcr.scripps.edu +started: 2026-11-19T22:53:55Z +exit: 0 +--- +pyxis: importing docker image: ghcr.io#brineylab/toolshed/tmol@sha256:97cdbaa7c71dfd694f15137702b15f5fcd2b3af5ef6df5edf49aaccb880d04b0 +pyxis: imported docker image: ghcr.io#brineylab/toolshed/tmol@sha256:97cdbaa7c71dfd694f15137702b15f5fcd2b3af5ef6df5edf49aaccb880d04b0 +node: gpu-swann-0 +GPU 0: NVIDIA L40S (UUID: GPU-ab31b94e-1404-7858-604e-4f2a6b62d33e) +Environment variable CCD_MIRROR_PATH not set. Will not be able to use function requiring this variable. To set it you may: + (1) add the line 'export VAR_NAME=path/to/variable' to your .bashrc or .zshrc file + (2) set it in your current shell with 'export VAR_NAME=path/to/variable' + (3) write it to a .env file in the root of the atomworks.io repository +Environment variable PDB_MIRROR_PATH not set. Will not be able to use function requiring this variable. To set it you may: + (1) add the line 'export VAR_NAME=path/to/variable' to your .bashrc or .zshrc file + (2) set it in your current shell with 'export VAR_NAME=path/to/variable' + (3) write it to a .env file in the root of the atomworks.io repository +tmol: scored 26 structure(s) on cuda in 8.8s -> /tmp/tmol.IDH8i3/scores.tsv +Environment variable CCD_MIRROR_PATH not set. Will not be able to use function requiring this variable. To set it you may: + (1) add the line 'export VAR_NAME=path/to/variable' to your .bashrc or .zshrc file + (2) set it in your current shell with 'export VAR_NAME=path/to/variable' + (3) write it to a .env file in the root of the atomworks.io repository +Environment variable PDB_MIRROR_PATH not set. Will not be able to use function requiring this variable. To set it you may: + (1) add the line 'export VAR_NAME=path/to/variable' to your .bashrc or .zshrc file + (2) set it in your current shell with 'export VAR_NAME=path/to/variable' + (3) write it to a .env file in the root of the atomworks.io repository +tmol: minimized 26 structure(s) on cuda: load 8.2s, minimize 3.4s, write 0.2s, score 3.1s +Environment variable CCD_MIRROR_PATH not set. Will not be able to use function requiring this variable. To set it you may: + (1) add the line 'export VAR_NAME=path/to/variable' to your .bashrc or .zshrc file + (2) set it in your current shell with 'export VAR_NAME=path/to/variable' + (3) write it to a .env file in the root of the atomworks.io repository +Environment variable PDB_MIRROR_PATH not set. Will not be able to use function requiring this variable. To set it you may: + (1) add the line 'export VAR_NAME=path/to/variable' to your .bashrc or .zshrc file + (2) set it in your current shell with 'export VAR_NAME=path/to/variable' + (3) write it to a .env file in the root of the atomworks.io repository +tmol: packed 26 structure(s) on cuda: load 8.3s, pack 1.1s, write 0.2s, score 3.3s +Environment variable CCD_MIRROR_PATH not set. Will not be able to use function requiring this variable. To set it you may: + (1) add the line 'export VAR_NAME=path/to/variable' to your .bashrc or .zshrc file + (2) set it in your current shell with 'export VAR_NAME=path/to/variable' + (3) write it to a .env file in the root of the atomworks.io repository +Environment variable PDB_MIRROR_PATH not set. Will not be able to use function requiring this variable. To set it you may: + (1) add the line 'export VAR_NAME=path/to/variable' to your .bashrc or .zshrc file + (2) set it in your current shell with 'export VAR_NAME=path/to/variable' + (3) write it to a .env file in the root of the atomworks.io repository +tmol: packed 26 structure(s) on cuda: load 8.3s, pack 1.1s, write 0.2s, score 3.2s +Environment variable CCD_MIRROR_PATH not set. Will not be able to use function requiring this variable. To set it you may: + (1) add the line 'export VAR_NAME=path/to/variable' to your .bashrc or .zshrc file + (2) set it in your current shell with 'export VAR_NAME=path/to/variable' + (3) write it to a .env file in the root of the atomworks.io repository +Environment variable PDB_MIRROR_PATH not set. Will not be able to use function requiring this variable. To set it you may: + (1) add the line 'export VAR_NAME=path/to/variable' to your .bashrc or .zshrc file + (2) set it in your current shell with 'export VAR_NAME=path/to/variable' + (3) write it to a .env file in the root of the atomworks.io repository +tmol: relaxed 26 structure(s) on cuda: load 8.2s, relax 16.1s, write 0.2s, score 3.3s +Environment variable CCD_MIRROR_PATH not set. Will not be able to use function requiring this variable. To set it you may: + (1) add the line 'export VAR_NAME=path/to/variable' to your .bashrc or .zshrc file + (2) set it in your current shell with 'export VAR_NAME=path/to/variable' + (3) write it to a .env file in the root of the atomworks.io repository +Environment variable PDB_MIRROR_PATH not set. Will not be able to use function requiring this variable. To set it you may: + (1) add the line 'export VAR_NAME=path/to/variable' to your .bashrc or .zshrc file + (2) set it in your current shell with 'export VAR_NAME=path/to/variable' + (3) write it to a .env file in the root of the atomworks.io repository +tmol: scored 26 structure(s) on cuda in 9.3s -> /tmp/tmol.IDH8i3/score-min.tsv +Environment variable CCD_MIRROR_PATH not set. Will not be able to use function requiring this variable. To set it you may: + (1) add the line 'export VAR_NAME=path/to/variable' to your .bashrc or .zshrc file + (2) set it in your current shell with 'export VAR_NAME=path/to/variable' + (3) write it to a .env file in the root of the atomworks.io repository +Environment variable PDB_MIRROR_PATH not set. Will not be able to use function requiring this variable. To set it you may: + (1) add the line 'export VAR_NAME=path/to/variable' to your .bashrc or .zshrc file + (2) set it in your current shell with 'export VAR_NAME=path/to/variable' + (3) write it to a .env file in the root of the atomworks.io repository +tmol test OK: 26 poses of 1UBQ at 195.065; minimize/pack/relax each lowered it; the seeded pair reproduces; scores.tsv chains through tmol score diff --git a/tools/tmol/Containerfile b/tools/tmol/Containerfile new file mode 100644 index 0000000..37b1414 --- /dev/null +++ b/tools/tmol/Containerfile @@ -0,0 +1,31 @@ +FROM docker.io/pytorch/pytorch:2.14.1-cuda12.6-cudnn9-runtime +ENV NVIDIA_VISIBLE_DEVICES=all NVIDIA_DRIVER_CAPABILITIES=compute,utility +# Pinned by version tag, not digest, on purpose: Docker Hub serves +# pytorch/pytorch's manifests by digest only intermittently -- a build +# pinned to this tag's own digest, its manifest's true sha256 +# (sha256:1262abc856a428933c0477b4be87079173a0b087b7691c47fcb396b7e5775c56, +# confirmed against the registry), failed at the pull with +# "manifest unknown" -- while the tag resolves every time, the way +# tools/ligandmpnn pins its pytorch base. The tag is pytorchbot's +# one-time push: torch 2.14.1+cu126, CUDA 12.6.3, Ubuntu 24.04 with +# python 3.12 at /usr/local, where pip sees the installed torch. +# tmol is a library with no command-line program of its own, so the image +# carries the shed's wrapper (bin/tmol: score, minimize, pack, relax, with +# defaults; past the defaults is the library way, python3 -I script.py). +# The base's torch 2.14.1+cu126 is what tmol-ops-cu12 1.0.0 is built for; +# pip sees it already installed and leaves it. The cu12 extra adds tmol's +# CUDA operators; the ligand extra adds Open Babel, which ligand +# preparation uses. tmol's beta2016 parameters ship inside the wheel, so +# nothing is fetched at build or run time and the image needs no offline +# flags. Runs on CPU too. +RUN pip install --no-cache-dir --break-system-packages "tmol[cu12,ligand]==1.0.0" +# Proven at build time, on the CPU: tmol imports, the wheel's parameter +# database (the beta2016 weights) loads, and the score function renders. +RUN python3 -I -c "import tmol, torch; from tmol import beta2016_score_function; sfxn = beta2016_score_function(torch.device('cpu')); print('tmol', tmol.__version__, '/ torch', torch.__version__, '/ cuda operators', tmol.has_cuda(), '/ beta2016 score types', len(sfxn.all_score_types()))" +# The upstream BSD-3-Clause license. +COPY LICENSE /opt/tmol/LICENSE +# The wrapper: every import it makes must resolve and its CLI wiring must +# hold, in this image, before it is pushed. +COPY --chmod=755 bin/tmol /usr/local/bin/tmol +RUN tmol --help > /dev/null && tmol pack --help > /dev/null +COPY test /toolshed/test diff --git a/tools/tmol/LICENSE b/tools/tmol/LICENSE new file mode 100644 index 0000000..9395155 --- /dev/null +++ b/tools/tmol/LICENSE @@ -0,0 +1,29 @@ +BSD 3-Clause License + +Copyright (c) 2026, Institute for Protein Design, University of Washington +All rights reserved. + +Redistribution and use in source and binary forms, with or without +modification, are permitted provided that the following conditions are met: + +1. Redistributions of source code must retain the above copyright notice, this + list of conditions and the following disclaimer. + +2. Redistributions in binary form must reproduce the above copyright notice, + this list of conditions and the following disclaimer in the documentation + and/or other materials provided with the distribution. + +3. Neither the name of the copyright holder nor the names of its + contributors may be used to endorse or promote products derived from + this software without specific prior written permission. + +THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS" +AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE +IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE +DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR CONTRIBUTORS BE LIABLE +FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL +DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR +SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER +CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, +OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE +OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. diff --git a/tools/tmol/SKILL.md b/tools/tmol/SKILL.md new file mode 100644 index 0000000..8c6fb02 --- /dev/null +++ b/tools/tmol/SKILL.md @@ -0,0 +1,202 @@ +--- +name: tmol +description: "tmol, RosettaCommons' GPU implementation of the Rosetta energy function in PyTorch: score, Cartesian-minimize, repack, and FastRelax protein and ligand structures, one file or a whole directory batched on one GPU. Reach for it to rescore or refine structures with the beta2016 energy; anything past the four verbs' defaults — per-term scores, movement masks, restraints, gradients — is a short python3 -I script against the same library, documented below." +metadata: + version: 1.0.0 +--- +# tmol + +Image: `ghcr.io/brineylab/toolshed/tmol:1.0.0` + +tmol 1.0.0 — the RosettaCommons `tmol` package, BSD-3-Clause (Institute for Protein Design; the upstream `LICENSE` ships in the image at `/opt/tmol/LICENSE`; documentation, quickstart, and runnable tutorials at https://github.com/RosettaCommons/tmol/tree/v1.0.0) — on the shed's CUDA-12.6 PyTorch 2.14.1 image (Python 3.12), with the `cu12` extra (`tmol-ops-cu12`, the CUDA operators) and the `ligand` extra (Open Babel, which ligand preparation uses). The beta2016 parameters ship inside the wheel: no weights are fetched, the image needs no network, and it runs on CPU too (`--device cpu`), slower there (measured below). tmol's scores are tmol score units — not kcal/mol, and not numerically interchangeable with Rosetta's. Upstream ships no command-line program, so the image carries the shed's wrapper, `tmol`, four verbs with defaults; everything past the defaults is the library way, `python3 -I script.py`, below. + +## Commands + +The four verbs. `` is one structure file or a directory of them (PDB, mmCIF, MOL2, or SDF, optionally `.gz`/`.zst` compressed); a directory's files load into one batch, and everything is processed together on the device. + +```bash +tmol score in/ scores.tsv +tmol minimize in/ out/ +tmol pack in/ out/ --seed 1 +tmol relax in/ out/ +``` + +- `score` writes a TSV, one row per input — `name`, `total`, then one column per weighted beta2016 term, `%.6f` values. Every verb scores with beta2016. +- `minimize` is Cartesian minimization with every atom free; `pack` is fixed-sequence side-chain repacking; `relax` is FastRelax with its defaults (Cartesian, every atom free). +- The optimizer verbs write one structure per input into `out/` — the input's own format and file name, compression stripped — plus `out/scores.tsv`, the same numbers `tmol score out/` reports again to within GPU reduction noise (measured ≤ 2×10⁻⁴ on the total; not byte-identical), so the verbs chain through files: + +```bash +tmol relax in/ out/ +tmol score out/ scores.tsv +``` + +- Every verb takes `--device DEV` (default: cuda when one is visible, else cpu); `pack` and `relax` also take `--seed N` (default 0). That is the whole surface: no other flags, no other score function, and the wrapper sequences nothing. + +The second way in is the library itself, under `python3 -I` — the same calls the wrapper makes. A script that loads a structure, relaxes it, and scores it again, all in one process (`pose.pdb` in, `relaxed.pdb` out): + +```bash +python3 -I relax_score.py pose.pdb relaxed.pdb +``` + +`relax_score.py`: + +```python +import sys + +import torch +from tmol import ( + beta2016_score_function, + fast_relax, + file_from_pose_stack, + pose_stack_from_file, +) + +source, output = sys.argv[1], sys.argv[2] +device = torch.device("cuda" if torch.cuda.is_available() else "cpu") +torch.manual_seed(0) + +pose = pose_stack_from_file(source, device, packer_seed=0) +sfxn = beta2016_score_function(device) +scorer = sfxn.render_whole_pose_scoring_module(pose) +with torch.no_grad(): + before = scorer(pose.coords).tolist() + +relaxed = fast_relax(pose, sfxn) + +# Re-render the scorer on the relaxed pose: its atom layout can change, +# and the old scorer is bound to the old layout. +scorer = sfxn.render_whole_pose_scoring_module(relaxed) +with torch.no_grad(): + after = scorer(relaxed.coords).tolist() + +file_from_pose_stack(relaxed, output) +print(f"before {before[0]:.6f}") +print(f"after {after[0]:.6f}") +``` + +Past the verbs' defaults, the pieces worth knowing: + +**Loading and batching.** `pose_stack_from_file(path, device, packer_seed=0)` loads one file into a one-pose stack. Several files that belong together — a directory, a protein and its ligands — load through one shared context, which is also how the wrapper loads a directory: + +```python +import biotite.structure +from tmol import PoseStackBuilder, atom_array_from_file, pose_stack_from_atom_array +from tmol.io import build_context_from_atom_array + +arrays = [atom_array_from_file(str(path)) for path in paths] +context = build_context_from_atom_array( + biotite.structure.concatenate(arrays), device, + prepare_ligands=True, ligand_seed=0, +) +poses = [ + pose_stack_from_atom_array(array, device, context=context, packer_seed=0) + for array in arrays +] +stack = PoseStackBuilder.from_poses(poses, device) +``` + +`prepare_ligands=True` parameterizes whatever noncanonical chemistry the inputs hold — small molecules, metals, nucleic acids — and is how protein–ligand complexes score. The score function must then be built from the context's database: `beta2016_score_function(device, param_db=context.parameter_database)`. Protein-only inputs can instead use plain `pose_stack_from_file` with `beta2016_score_function(device)`, which share tmol's default database. Waters (HOH) are dropped on load, in every path. + +**Per-term scores.** + +```python +scorer = sfxn.render_whole_pose_scoring_module(stack) +with torch.no_grad(): + totals = scorer(stack.coords) # [n_poses] + terms = scorer(stack.coords, sum_terms=False) # [n_terms, n_poses] +names = [score_type.name for score_type in sfxn.all_score_types()] +``` + +`sum_terms=False` returns weighted terms; add `apply_weights=False` for the unweighted decomposition. `render_block_pair_scoring_module()` attributes the score block by block. Render a scorer again after anything that can change the stack's atom layout (packing, relaxation); a coordinate-only update with the same layout reuses it. + +**Minimization.** `run_cart_min(stack, sfxn)` optimizes coordinates with every atom free (the `minimize` verb). Pass a boolean mask with `stack.coords`' shape to move only chosen atoms. Torsions and rigid-body jumps are `run_kin_min`'s territory, with a configured `FoldForest` and `MoveMap` — a `MoveMap` enables nothing by existing; set its flags. + +**Packing.** Fixed-sequence repacking, the `pack` verb's recipe: + +```python +from tmol import PackerPalette, PackerTask, pack_rotamers +from tmol.pack import ( + FixedAAChiSampler, + IncludeCurrentSampler, + create_dunbrack_sampler_from_database, +) + +task = PackerTask(stack, PackerPalette()) +task.restrict_to_repacking() +task.add_conformer_sampler( + create_dunbrack_sampler_from_database(context.parameter_database, device) +) +task.add_conformer_sampler(FixedAAChiSampler()) +task.add_conformer_sampler(IncludeCurrentSampler()) +packed = pack_rotamers(stack, sfxn, task, seed=1) +``` + +`restrict_to_repacking()` keeps the sequence; design needs an explicit allowed-identity task. `disable_packing_by_block_mask()` freezes chosen blocks. `pack_rotamers` chunks large stacks itself and its `seed=` forks torch's generator; render a new scorer on the packed stack. + +**FastRelax.** `relax` is `fast_relax(stack, sfxn)` — Cartesian, every atom free. The `minimizer=` argument carries the mode and its settings: + +```python +from tmol import KinematicRelaxMinimizer, fast_relax + +torsions = fast_relax(stack, sfxn, minimizer=KinematicRelaxMinimizer()) +``` + +`CartesianRelaxMinimizer(cartesian_move_map)` and `KinematicRelaxMinimizer(move_map, fold_forest)` each reject the other's move-map type; without a move map the kinematic minimizer frees every torsion and jump, while `MoveMap.from_pose_stack` frees nothing until enabled. `cuda_graph=False` forces eager execution. `fast_relax` takes no seed argument — `torch.manual_seed(N)` right before it, which is what the wrapper does. + +**Constraints.** A constraint must be attached to the pose and its score term must carry a nonzero weight in the score function; miss either and it is silently inactive. + +**Coordinate gradients.** The score is differentiable in the coordinates: + +```python +coords = stack.coords.detach().clone().requires_grad_(True) +total = scorer(coords).sum() +total.backward() +gradient = coords.grad # stack.coords' shape +``` + +## Inputs and outputs + +- **In:** one structure file, or a directory of them (read flat, not recursively): PDB (`.pdb`, `.ent`), mmCIF (`.cif`, `.mmcif`, `.bcif`), MOL2, SDF, or MDL `.mol`, each optionally `.gz`, `.gzip`, or `.zst`. Proteins, nucleic acids, ligands, and complexes; waters are dropped. +- **Out:** `score` writes its TSV with a `name` column (the input's file name), `total`, and one column per weighted beta2016 term. The optimizer verbs write one structure per input — same format, same file name, compression stripped — plus `scores.tsv` in the same shape, which `tmol score ` reproduces to within GPU reduction noise (≤ 2×10⁻⁴ measured on the total). + +## Compute + +One GPU. Measured end to end — load, compute, write, and the written structures' rescore — on the test's fixture (26 poses of 1UBQ, 72 KB a pose) on one NVIDIA L40S (46 GB, driver 580.178.04), torch 2.14.1+cu126: + +| verb | time | rate | +| --- | --- | --- | +| `score` | 23–24 s | ~1.1 structures/s | +| `minimize` | 28 s | ~0.9 structures/s | +| `pack` | 16–18 s | ~1.5 structures/s | +| `relax` | 38–39 s | ~0.7 structures/s | + +The library way's one-pose relax-then-score script took 14 s. Most of a small run is fixed cost — structure load, parameter-database rendering, first-use CUDA compilation — so rates rise with batch size until the packer's chunking (25 poses a chunk here) and GPU memory bind; every verb prints its own load / operate / write / score breakdown to stderr. On CPU (16 cluster cores) the same batch ran `score` 29 s, `minimize` 29 s, `pack` 21 s, `relax` 72 s — 1.0–1.9× slower end to end, the gap widening with the compute's share — and the one-pose script 77 s (5.6× slower: a single pose's tensors leave the CPU cores idle where a batch spreads over them). The fixture's CPU and GPU totals agree within the same ~3×10⁻⁴ reduction noise. + +## Running it + +On a laptop, from the directory holding `in/` (the wrapper picks the GPU when one is visible, else CPU): + +```bash +podman run --rm --device nvidia.com/gpu=all -v "$PWD":/work -w /work \ + ghcr.io/brineylab/toolshed/tmol:1.0.0 tmol score in/ scores.tsv +``` + +On a Slurm cluster with pyxis: + +```bash +srun --gres=gpu:1 --container-image=ghcr.io#brineylab/toolshed/tmol:1.0.0 \ + --container-mounts="$PWD":/work --container-workdir=/work \ + tmol relax in/ out/ +``` + +## Gotchas + +- The test's fixture is 26 copies of water-stripped 1UBQ because tmol's packer chunks a pose stack: a pose of at most 256 blocks (1UBQ's 76) packs in chunks of 25 poses, so 26 copies crosses the boundary — the test exercises the chunked path (25 poses, then 1); fewer than 25 poses never leaves one chunk. `TMOL_PACK_MAX_POSES_PER_CHUNK` lowers the bound when GPU memory is tight. +- tmol's scores are tmol score units, not kcal/mol, and not numerically interchangeable with Rosetta's; compare tmol numbers with tmol numbers. +- Waters (HOH) are silently dropped on load, in every path; everything else non-protein — ligands, metals, DNA — is parameterized and kept. +- Seeding: every verb seeds torch and the load-time hydrogen packer; `--seed` (pack/relax, default 0) additionally drives the packer's annealer and FastRelax. `score` always runs at seed 0, and the optimizer verbs score their written outputs at seed 0, which is why `tmol score ` reproduces `scores.tsv` within the reduction noise above. `fast_relax` takes no seed argument — library callers `torch.manual_seed` before it. +- Seeded reproducibility, measured on the fixture: two `tmol pack --seed 1` runs wrote byte-identical structures (their `scores.tsv` totals 1.6×10⁻⁴ apart — the reduction noise), but two `tmol relax --seed 1` runs did **not**: CUDA score reductions are non-deterministic at the ~10⁻⁴ level, and FastRelax's iterated pack-and-minimize cycles amplify that noise into different local minima — the pair's totals differed by up to 3.1 and a few side-chain atoms by up to 5.4 Å, pose for pose, at the same seed. Treat `relax` as stochastic: compare energies, not coordinates, and take the best of several seeds when the structure matters. +- The image is CUDA 12.6: it needs an NVIDIA driver from the CUDA 12 era (≥ 525.60.13). The campaign measured on an L40S with driver 580.178.04. +- On CPU (`--device cpu`) every verb works and the totals agree with the GPU's within the same ~3×10⁻⁴ noise, but the compute falls behind as the batch shrinks: 1.0–1.9× slower per verb on the 26-pose fixture and 5.6× on a single pose (see Compute). +- The wrapper runs under `python3 -I`, and the library way above is the same flag: a caller's `PYTHONPATH` or user site-packages never replace what the image tested. +- A directory is read flat (no recursion) and only structure-suffixed files are collected, so the `scores.tsv` inside an optimizer verb's outdir does not confuse a later `tmol score` of that directory. diff --git a/tools/tmol/bin/tmol b/tools/tmol/bin/tmol new file mode 100755 index 0000000..dec07f6 --- /dev/null +++ b/tools/tmol/bin/tmol @@ -0,0 +1,330 @@ +#!/usr/bin/env -S python3 -I +"""tmol: score, minimize, pack, and relax structures with tmol. + +tmol (https://github.com/RosettaCommons/tmol, BSD-3-Clause) is +RosettaCommons' GPU implementation of the Rosetta energy function in +PyTorch. This wrapper gives it four verbs with defaults; anything +past the defaults is the library way, ``python3 -I script.py``, +documented in this tool's skill. + + tmol score [--device DEV] + tmol minimize [--device DEV] + tmol pack [--seed N] [--device DEV] + tmol relax [--seed N] [--device DEV] + +```` is one structure file (PDB, mmCIF, MOL2 or SDF, optionally +gzip/Zstandard compressed) or a directory of them; every input is +loaded into one batch and processed together on the device. All verbs +score with the beta2016 score function. Each verb seeds torch and the +load-time hydrogen packer, and ``--seed`` (default 0) additionally +drives packing and relaxation: seeded ``pack`` runs write +byte-identical structures, while seeded ``relax`` runs on GPU land in +different local minima (see this tool's skill). The optimizer verbs +write one structure per input into ```` (the input's format +and file name) plus ``/scores.tsv``, which matches what +``tmol score `` reports to within GPU reduction noise (~2e-4), +so the verbs chain through files: ``tmol relax in/ out/`` then +``tmol score out/ s.tsv``. +""" + +import argparse +import sys +import time +from pathlib import Path + +import biotite.structure +import torch +from tmol import ( + PackerPalette, + PackerTask, + PoseStackBuilder, + atom_array_from_file, + beta2016_score_function, + fast_relax, + file_from_pose_stack, + pack_rotamers, + pose_stack_from_atom_array, + run_cart_min, +) +from tmol.io import build_context_from_atom_array +from tmol.pack import ( + FixedAAChiSampler, + IncludeCurrentSampler, + create_dunbrack_sampler_from_database, +) + +_STRUCTURE_SUFFIXES = { + ".pdb", + ".ent", + ".cif", + ".mmcif", + ".bcif", + ".mol2", + ".sdf", + ".mol", +} +_COMPRESSION_SUFFIXES = {".gz", ".gzip", ".zst"} + +# `tmol score` always runs at this seed, and the optimizer verbs score +# their written outputs at it too, so scores.tsv matches a later +# `tmol score ` up to GPU reduction noise (~2e-4). +_SCORE_SEED = 0 + +_PAST_TENSE = {"minimize": "minimized", "pack": "packed", "relax": "relaxed"} + + +def _note(message): + print(f"tmol: {message}", file=sys.stderr, flush=True) + + +def _suffixes(path): + """A path's suffixes in path order, lowercased (the extension is last).""" + return [suffix.lower() for suffix in Path(path).suffixes] + + +def _is_structure_file(path): + suffixes = _suffixes(path) + if not suffixes: + return False + if suffixes[-1] in _COMPRESSION_SUFFIXES: + # e.g. 1UBQ.pdb.gz: a compressed structure file + return len(suffixes) > 1 and suffixes[-2] in _STRUCTURE_SUFFIXES + return suffixes[-1] in _STRUCTURE_SUFFIXES + + +def collect_inputs(input_path): + """The sorted list of structure files an input path names.""" + path = Path(input_path) + if path.is_file(): + return [path] + if path.is_dir(): + files = sorted( + entry + for entry in path.iterdir() + if entry.is_file() and _is_structure_file(entry) + ) + if not files: + raise SystemExit(f"tmol: no structure files found in {path}") + return files + raise SystemExit(f"tmol: input not found: {path}") + + +def output_path_for(outdir, input_file): + """Where one input's optimized structure is written.""" + name = Path(input_file).name + suffixes = _suffixes(name) + if suffixes and suffixes[-1] in _COMPRESSION_SUFFIXES: + # Compressed inputs are written back uncompressed. + name = Path(name).with_suffix("").name + return Path(outdir) / name + + +def resolve_device(value): + if value is None: + return torch.device("cuda" if torch.cuda.is_available() else "cpu") + try: + return torch.device(value) + except RuntimeError as error: + raise SystemExit(f"tmol: bad --device {value!r}: {error}") + + +def load_batch(files, device, seed): + """Load every input file into one pose stack, sharing one context. + + The build context (residue types, parameter database, packed block + types) is derived once from all inputs together -- preparing any + ligands they hold -- and each file's pose is built against it, so + mixed protein/ligand inputs land in one batch. This is tmol's + documented way to score many structures that share ligands. + """ + arrays = [atom_array_from_file(str(file)) for file in files] + combined = ( + arrays[0] + if len(arrays) == 1 + else biotite.structure.concatenate(arrays) + ) + context = build_context_from_atom_array( + combined, device, prepare_ligands=True, ligand_seed=seed + ) + poses = [ + pose_stack_from_atom_array(array, device, context=context, packer_seed=seed) + for array in arrays + ] + stack = ( + poses[0] if len(poses) == 1 else PoseStackBuilder.from_poses(poses, device) + ) + return stack, context + + +def score_stack(stack, context, device): + """Total and per-term beta2016 scores for every pose in a stack.""" + sfxn = beta2016_score_function(device, param_db=context.parameter_database) + scorer = sfxn.render_whole_pose_scoring_module(stack) + with torch.no_grad(): + totals = scorer(stack.coords).detach().cpu() + terms = scorer(stack.coords, sum_terms=False).detach().cpu() + names = [score_type.name for score_type in sfxn.all_score_types()] + if tuple(terms.shape) != (len(names), stack.n_poses): + raise SystemExit( + f"tmol: unexpected score-term shape {tuple(terms.shape)} " + f"for {len(names)} terms and {stack.n_poses} poses" + ) + return names, totals, terms + + +def write_scores_tsv(path, files, names, totals, terms): + path = Path(path) + path.parent.mkdir(parents=True, exist_ok=True) + with open(path, "w") as handle: + handle.write("\t".join(["name", "total", *names]) + "\n") + for pose, file in enumerate(files): + values = [totals[pose].item()] + values.extend(terms[:, pose].tolist()) + row = "\t".join([file.name, *(f"{value:.6f}" for value in values)]) + handle.write(row + "\n") + + +def score_files(files, device, seed): + """Load, score and tabulate a list of files (the score verb's core).""" + stack, context = load_batch(files, device, seed) + names, totals, terms = score_stack(stack, context, device) + return names, totals, terms + + +def verb_score(args): + device = resolve_device(args.device) + started = time.perf_counter() + torch.manual_seed(_SCORE_SEED) + files = collect_inputs(args.input) + names, totals, terms = score_files(files, device, _SCORE_SEED) + write_scores_tsv(args.scores_tsv, files, names, totals, terms) + _note( + f"scored {len(files)} structure(s) on {device} in " + f"{time.perf_counter() - started:.1f}s -> {args.scores_tsv}" + ) + + +def verb_optimize(args, operation, seed): + device = resolve_device(args.device) + started = time.perf_counter() + torch.manual_seed(seed) + files = collect_inputs(args.input) + stack, context = load_batch(files, device, seed) + sfxn = beta2016_score_function(device, param_db=context.parameter_database) + loaded = time.perf_counter() + + if operation == "minimize": + result = run_cart_min(stack, sfxn) + elif operation == "pack": + task = PackerTask(stack, PackerPalette()) + task.restrict_to_repacking() + task.add_conformer_sampler( + create_dunbrack_sampler_from_database( + context.parameter_database, device + ) + ) + task.add_conformer_sampler(FixedAAChiSampler()) + task.add_conformer_sampler(IncludeCurrentSampler()) + result = pack_rotamers(stack, sfxn, task, seed=seed) + elif operation == "relax": + # fast_relax has no seed argument; its packer draws from torch's + # global generator, so reseed right before the call. + torch.manual_seed(seed) + result = fast_relax(stack, sfxn) + else: + raise AssertionError(operation) + optimized = time.perf_counter() + + outdir = Path(args.outdir) + outdir.mkdir(parents=True, exist_ok=True) + outputs = [output_path_for(outdir, file) for file in files] + for pose, output in enumerate(outputs): + file_from_pose_stack(result.split(pose), str(output)) + written = time.perf_counter() + + # scores.tsv describes the written structures as `tmol score` sees + # them, so the verbs chain through files. + names, totals, terms = score_files(outputs, device, _SCORE_SEED) + write_scores_tsv(outdir / "scores.tsv", outputs, names, totals, terms) + _note( + f"{_PAST_TENSE[operation]} {len(files)} structure(s) on {device}: " + f"load {loaded - started:.1f}s, {operation} {optimized - loaded:.1f}s, " + f"write {written - optimized:.1f}s, score {time.perf_counter() - written:.1f}s" + ) + + +def build_parser(): + parser = argparse.ArgumentParser( + prog="tmol", + description=( + "Score, minimize, pack, or relax structures with tmol's " + "beta2016 score function. Everything past these defaults is " + "the library way: python3 -I script.py (see the tool's skill)." + ), + ) + subparsers = parser.add_subparsers(dest="verb", required=True) + + def add_verb(name, help_text): + subparser = subparsers.add_parser(name, help=help_text) + subparser.add_argument( + "input", + help="one structure file, or a directory of them", + ) + subparser.add_argument( + "--device", + metavar="DEV", + default=None, + help="torch device (default: cuda when available, else cpu)", + ) + return subparser + + score_parser = add_verb( + "score", "score structures and write a TSV of beta2016 totals and terms" + ) + score_parser.add_argument( + "scores_tsv", help="output TSV path (columns: name, total, one per term)" + ) + score_parser.set_defaults(run=verb_score) + + minimize_parser = add_verb( + "minimize", "Cartesian minimization with every atom free" + ) + minimize_parser.add_argument("outdir", help="output directory") + minimize_parser.set_defaults( + run=lambda args: verb_optimize(args, "minimize", _SCORE_SEED) + ) + + pack_parser = add_verb("pack", "fixed-sequence side-chain repacking") + pack_parser.add_argument("outdir", help="output directory") + pack_parser.add_argument( + "--seed", + type=int, + default=0, + help="random seed (default: 0)", + ) + pack_parser.set_defaults( + run=lambda args: verb_optimize(args, "pack", args.seed) + ) + + relax_parser = add_verb("relax", "fast_relax with its defaults") + relax_parser.add_argument("outdir", help="output directory") + relax_parser.add_argument( + "--seed", + type=int, + default=0, + help="random seed (default: 0)", + ) + relax_parser.set_defaults( + run=lambda args: verb_optimize(args, "relax", args.seed) + ) + + return parser + + +def main(argv=None): + args = build_parser().parse_args(argv) + args.run(args) + + +if __name__ == "__main__": + main() diff --git a/tools/tmol/test/1UBQ.pdb b/tools/tmol/test/1UBQ.pdb new file mode 100644 index 0000000..296f4d1 --- /dev/null +++ b/tools/tmol/test/1UBQ.pdb @@ -0,0 +1,911 @@ +HEADER CHROMOSOMAL PROTEIN 02-JAN-87 1UBQ +TITLE STRUCTURE OF UBIQUITIN REFINED AT 1.8 ANGSTROMS RESOLUTION +COMPND MOL_ID: 1; +COMPND 2 MOLECULE: UBIQUITIN; +COMPND 3 CHAIN: A; +COMPND 4 ENGINEERED: YES +SOURCE MOL_ID: 1; +SOURCE 2 ORGANISM_SCIENTIFIC: HOMO SAPIENS; +SOURCE 3 ORGANISM_COMMON: HUMAN; +SOURCE 4 ORGANISM_TAXID: 9606 +KEYWDS CHROMOSOMAL PROTEIN +EXPDTA X-RAY DIFFRACTION +AUTHOR S.VIJAY-KUMAR,C.E.BUGG,W.J.COOK +REVDAT 6 14-FEB-24 1UBQ 1 REMARK +REVDAT 5 09-MAR-11 1UBQ 1 REMARK +REVDAT 4 24-FEB-09 1UBQ 1 VERSN +REVDAT 3 01-APR-03 1UBQ 1 JRNL +REVDAT 2 16-JUL-87 1UBQ 1 JRNL REMARK +REVDAT 1 16-APR-87 1UBQ 0 +JRNL AUTH S.VIJAY-KUMAR,C.E.BUGG,W.J.COOK +JRNL TITL STRUCTURE OF UBIQUITIN REFINED AT 1.8 A RESOLUTION. +JRNL REF J.MOL.BIOL. V. 194 531 1987 +JRNL REFN ISSN 0022-2836 +JRNL PMID 3041007 +JRNL DOI 10.1016/0022-2836(87)90679-6 +REMARK 1 +REMARK 1 REFERENCE 1 +REMARK 1 AUTH S.VIJAY-KUMAR,C.E.BUGG,K.D.WILKINSON,R.D.VIERSTRA, +REMARK 1 AUTH 2 P.M.HATFIELD,W.J.COOK +REMARK 1 TITL COMPARISON OF THE THREE-DIMENSIONAL STRUCTURES OF HUMAN, +REMARK 1 TITL 2 YEAST, AND OAT UBIQUITIN +REMARK 1 REF J.BIOL.CHEM. V. 262 6396 1987 +REMARK 1 REFN ISSN 0021-9258 +REMARK 1 REFERENCE 2 +REMARK 1 AUTH S.VIJAY-KUMAR,C.E.BUGG,K.D.WILKINSON,W.J.COOK +REMARK 1 TITL THREE-DIMENSIONAL STRUCTURE OF UBIQUITIN AT 2.8 ANGSTROMS +REMARK 1 TITL 2 RESOLUTION +REMARK 1 REF PROC.NATL.ACAD.SCI.USA V. 82 3582 1985 +REMARK 1 REFN ISSN 0027-8424 +REMARK 1 REFERENCE 3 +REMARK 1 AUTH W.J.COOK,F.L.SUDDATH,C.E.BUGG,G.GOLDSTEIN +REMARK 1 TITL CRYSTALLIZATION AND PRELIMINARY X-RAY INVESTIGATION OF +REMARK 1 TITL 2 UBIQUITIN, A NON-HISTONE CHROMOSOMAL PROTEIN +REMARK 1 REF J.MOL.BIOL. V. 130 353 1979 +REMARK 1 REFN ISSN 0022-2836 +REMARK 1 REFERENCE 4 +REMARK 1 AUTH D.H.SCHLESINGER,G.GOLDSTEIN +REMARK 1 TITL MOLECULAR CONSERVATION OF 74 AMINO ACID SEQUENCE OF +REMARK 1 TITL 2 UBIQUITIN BETWEEN CATTLE AND MAN +REMARK 1 REF NATURE V. 255 423 1975 +REMARK 1 REFN ISSN 0028-0836 +REMARK 2 +REMARK 2 RESOLUTION. 1.80 ANGSTROMS. +REMARK 3 +REMARK 3 REFINEMENT. +REMARK 3 PROGRAM : PROLSQ +REMARK 3 AUTHORS : KONNERT,HENDRICKSON +REMARK 3 +REMARK 3 DATA USED IN REFINEMENT. +REMARK 3 RESOLUTION RANGE HIGH (ANGSTROMS) : 1.80 +REMARK 3 RESOLUTION RANGE LOW (ANGSTROMS) : NULL +REMARK 3 DATA CUTOFF (SIGMA(F)) : NULL +REMARK 3 COMPLETENESS FOR RANGE (%) : NULL +REMARK 3 NUMBER OF REFLECTIONS : NULL +REMARK 3 +REMARK 3 FIT TO DATA USED IN REFINEMENT. +REMARK 3 CROSS-VALIDATION METHOD : NULL +REMARK 3 FREE R VALUE TEST SET SELECTION : NULL +REMARK 3 R VALUE (WORKING + TEST SET) : 0.176 +REMARK 3 R VALUE (WORKING SET) : NULL +REMARK 3 FREE R VALUE : NULL +REMARK 3 FREE R VALUE TEST SET SIZE (%) : NULL +REMARK 3 FREE R VALUE TEST SET COUNT : NULL +REMARK 3 +REMARK 3 FIT/AGREEMENT OF MODEL WITH ALL DATA. +REMARK 3 R VALUE (WORKING + TEST SET, NO CUTOFF) : NULL +REMARK 3 R VALUE (WORKING SET, NO CUTOFF) : NULL +REMARK 3 FREE R VALUE (NO CUTOFF) : NULL +REMARK 3 FREE R VALUE TEST SET SIZE (%, NO CUTOFF) : NULL +REMARK 3 FREE R VALUE TEST SET COUNT (NO CUTOFF) : NULL +REMARK 3 TOTAL NUMBER OF REFLECTIONS (NO CUTOFF) : NULL +REMARK 3 +REMARK 3 NUMBER OF NON-HYDROGEN ATOMS USED IN REFINEMENT. +REMARK 3 PROTEIN ATOMS : 602 +REMARK 3 NUCLEIC ACID ATOMS : 0 +REMARK 3 HETEROGEN ATOMS : 0 +REMARK 3 SOLVENT ATOMS : 58 +REMARK 3 +REMARK 3 B VALUES. +REMARK 3 FROM WILSON PLOT (A**2) : NULL +REMARK 3 MEAN B VALUE (OVERALL, A**2) : NULL +REMARK 3 OVERALL ANISOTROPIC B VALUE. +REMARK 3 B11 (A**2) : NULL +REMARK 3 B22 (A**2) : NULL +REMARK 3 B33 (A**2) : NULL +REMARK 3 B12 (A**2) : NULL +REMARK 3 B13 (A**2) : NULL +REMARK 3 B23 (A**2) : NULL +REMARK 3 +REMARK 3 ESTIMATED COORDINATE ERROR. +REMARK 3 ESD FROM LUZZATI PLOT (A) : NULL +REMARK 3 ESD FROM SIGMAA (A) : NULL +REMARK 3 LOW RESOLUTION CUTOFF (A) : NULL +REMARK 3 +REMARK 3 RMS DEVIATIONS FROM IDEAL VALUES. +REMARK 3 DISTANCE RESTRAINTS. RMS SIGMA +REMARK 3 BOND LENGTH (A) : 0.016 ; NULL +REMARK 3 ANGLE DISTANCE (A) : NULL ; NULL +REMARK 3 INTRAPLANAR 1-4 DISTANCE (A) : NULL ; NULL +REMARK 3 H-BOND OR METAL COORDINATION (A) : NULL ; NULL +REMARK 3 +REMARK 3 PLANE RESTRAINT (A) : NULL ; NULL +REMARK 3 CHIRAL-CENTER RESTRAINT (A**3) : NULL ; NULL +REMARK 3 +REMARK 3 NON-BONDED CONTACT RESTRAINTS. +REMARK 3 SINGLE TORSION (A) : NULL ; NULL +REMARK 3 MULTIPLE TORSION (A) : NULL ; NULL +REMARK 3 H-BOND (X...Y) (A) : NULL ; NULL +REMARK 3 H-BOND (X-H...Y) (A) : NULL ; NULL +REMARK 3 +REMARK 3 CONFORMATIONAL TORSION ANGLE RESTRAINTS. +REMARK 3 SPECIFIED (DEGREES) : NULL ; NULL +REMARK 3 PLANAR (DEGREES) : NULL ; NULL +REMARK 3 STAGGERED (DEGREES) : NULL ; NULL +REMARK 3 TRANSVERSE (DEGREES) : NULL ; NULL +REMARK 3 +REMARK 3 ISOTROPIC THERMAL FACTOR RESTRAINTS. RMS SIGMA +REMARK 3 MAIN-CHAIN BOND (A**2) : NULL ; NULL +REMARK 3 MAIN-CHAIN ANGLE (A**2) : NULL ; NULL +REMARK 3 SIDE-CHAIN BOND (A**2) : NULL ; NULL +REMARK 3 SIDE-CHAIN ANGLE (A**2) : NULL ; NULL +REMARK 3 +REMARK 3 OTHER REFINEMENT REMARKS: NULL +REMARK 4 +REMARK 4 1UBQ COMPLIES WITH FORMAT V. 3.30, 13-JUL-11 +REMARK 100 +REMARK 100 THIS ENTRY HAS BEEN PROCESSED BY BNL. +REMARK 100 THE DEPOSITION ID IS D_1000176905. +REMARK 200 +REMARK 200 EXPERIMENTAL DETAILS +REMARK 200 EXPERIMENT TYPE : X-RAY DIFFRACTION +REMARK 200 DATE OF DATA COLLECTION : NULL +REMARK 200 TEMPERATURE (KELVIN) : NULL +REMARK 200 PH : NULL +REMARK 200 NUMBER OF CRYSTALS USED : NULL +REMARK 200 +REMARK 200 SYNCHROTRON (Y/N) : NULL +REMARK 200 RADIATION SOURCE : NULL +REMARK 200 BEAMLINE : NULL +REMARK 200 X-RAY GENERATOR MODEL : NULL +REMARK 200 MONOCHROMATIC OR LAUE (M/L) : NULL +REMARK 200 WAVELENGTH OR RANGE (A) : NULL +REMARK 200 MONOCHROMATOR : NULL +REMARK 200 OPTICS : NULL +REMARK 200 +REMARK 200 DETECTOR TYPE : NULL +REMARK 200 DETECTOR MANUFACTURER : NULL +REMARK 200 INTENSITY-INTEGRATION SOFTWARE : NULL +REMARK 200 DATA SCALING SOFTWARE : NULL +REMARK 200 +REMARK 200 NUMBER OF UNIQUE REFLECTIONS : NULL +REMARK 200 RESOLUTION RANGE HIGH (A) : NULL +REMARK 200 RESOLUTION RANGE LOW (A) : NULL +REMARK 200 REJECTION CRITERIA (SIGMA(I)) : NULL +REMARK 200 +REMARK 200 OVERALL. +REMARK 200 COMPLETENESS FOR RANGE (%) : NULL +REMARK 200 DATA REDUNDANCY : NULL +REMARK 200 R MERGE (I) : NULL +REMARK 200 R SYM (I) : NULL +REMARK 200 FOR THE DATA SET : NULL +REMARK 200 +REMARK 200 IN THE HIGHEST RESOLUTION SHELL. +REMARK 200 HIGHEST RESOLUTION SHELL, RANGE HIGH (A) : NULL +REMARK 200 HIGHEST RESOLUTION SHELL, RANGE LOW (A) : NULL +REMARK 200 COMPLETENESS FOR SHELL (%) : NULL +REMARK 200 DATA REDUNDANCY IN SHELL : NULL +REMARK 200 R MERGE FOR SHELL (I) : NULL +REMARK 200 R SYM FOR SHELL (I) : NULL +REMARK 200 FOR SHELL : NULL +REMARK 200 +REMARK 200 DIFFRACTION PROTOCOL: NULL +REMARK 200 METHOD USED TO DETERMINE THE STRUCTURE: NULL +REMARK 200 SOFTWARE USED: NULL +REMARK 200 STARTING MODEL: NULL +REMARK 200 +REMARK 200 REMARK: NULL +REMARK 280 +REMARK 280 CRYSTAL +REMARK 280 SOLVENT CONTENT, VS (%): 32.94 +REMARK 280 MATTHEWS COEFFICIENT, VM (ANGSTROMS**3/DA): 1.83 +REMARK 280 +REMARK 280 CRYSTALLIZATION CONDITIONS: NULL +REMARK 290 +REMARK 290 CRYSTALLOGRAPHIC SYMMETRY +REMARK 290 SYMMETRY OPERATORS FOR SPACE GROUP: P 21 21 21 +REMARK 290 +REMARK 290 SYMOP SYMMETRY +REMARK 290 NNNMMM OPERATOR +REMARK 290 1555 X,Y,Z +REMARK 290 2555 -X+1/2,-Y,Z+1/2 +REMARK 290 3555 -X,Y+1/2,-Z+1/2 +REMARK 290 4555 X+1/2,-Y+1/2,-Z +REMARK 290 +REMARK 290 WHERE NNN -> OPERATOR NUMBER +REMARK 290 MMM -> TRANSLATION VECTOR +REMARK 290 +REMARK 290 CRYSTALLOGRAPHIC SYMMETRY TRANSFORMATIONS +REMARK 290 THE FOLLOWING TRANSFORMATIONS OPERATE ON THE ATOM/HETATM +REMARK 290 RECORDS IN THIS ENTRY TO PRODUCE CRYSTALLOGRAPHICALLY +REMARK 290 RELATED MOLECULES. +REMARK 290 SMTRY1 1 1.000000 0.000000 0.000000 0.00000 +REMARK 290 SMTRY2 1 0.000000 1.000000 0.000000 0.00000 +REMARK 290 SMTRY3 1 0.000000 0.000000 1.000000 0.00000 +REMARK 290 SMTRY1 2 -1.000000 0.000000 0.000000 25.42000 +REMARK 290 SMTRY2 2 0.000000 -1.000000 0.000000 0.00000 +REMARK 290 SMTRY3 2 0.000000 0.000000 1.000000 14.47500 +REMARK 290 SMTRY1 3 -1.000000 0.000000 0.000000 0.00000 +REMARK 290 SMTRY2 3 0.000000 1.000000 0.000000 21.38500 +REMARK 290 SMTRY3 3 0.000000 0.000000 -1.000000 14.47500 +REMARK 290 SMTRY1 4 1.000000 0.000000 0.000000 25.42000 +REMARK 290 SMTRY2 4 0.000000 -1.000000 0.000000 21.38500 +REMARK 290 SMTRY3 4 0.000000 0.000000 -1.000000 0.00000 +REMARK 290 +REMARK 290 REMARK: NULL +REMARK 300 +REMARK 300 BIOMOLECULE: 1 +REMARK 300 SEE REMARK 350 FOR THE AUTHOR PROVIDED AND/OR PROGRAM +REMARK 300 GENERATED ASSEMBLY INFORMATION FOR THE STRUCTURE IN +REMARK 300 THIS ENTRY. THE REMARK MAY ALSO PROVIDE INFORMATION ON +REMARK 300 BURIED SURFACE AREA. +REMARK 350 +REMARK 350 COORDINATES FOR A COMPLETE MULTIMER REPRESENTING THE KNOWN +REMARK 350 BIOLOGICALLY SIGNIFICANT OLIGOMERIZATION STATE OF THE +REMARK 350 MOLECULE CAN BE GENERATED BY APPLYING BIOMT TRANSFORMATIONS +REMARK 350 GIVEN BELOW. BOTH NON-CRYSTALLOGRAPHIC AND +REMARK 350 CRYSTALLOGRAPHIC OPERATIONS ARE GIVEN. +REMARK 350 +REMARK 350 BIOMOLECULE: 1 +REMARK 350 AUTHOR DETERMINED BIOLOGICAL UNIT: MONOMERIC +REMARK 350 APPLY THE FOLLOWING TO CHAINS: A +REMARK 350 BIOMT1 1 1.000000 0.000000 0.000000 0.00000 +REMARK 350 BIOMT2 1 0.000000 1.000000 0.000000 0.00000 +REMARK 350 BIOMT3 1 0.000000 0.000000 1.000000 0.00000 +REMARK 500 +REMARK 500 GEOMETRY AND STEREOCHEMISTRY +REMARK 500 SUBTOPIC: CLOSE CONTACTS +REMARK 500 +REMARK 500 THE FOLLOWING ATOMS THAT ARE RELATED BY CRYSTALLOGRAPHIC +REMARK 500 SYMMETRY ARE IN CLOSE CONTACT. AN ATOM LOCATED WITHIN 0.15 +REMARK 500 ANGSTROMS OF A SYMMETRY RELATED ATOM IS ASSUMED TO BE ON A +REMARK 500 SPECIAL POSITION AND IS, THEREFORE, LISTED IN REMARK 375 +REMARK 500 INSTEAD OF REMARK 500. ATOMS WITH NON-BLANK ALTERNATE +REMARK 500 LOCATION INDICATORS ARE NOT INCLUDED IN THE CALCULATIONS. +REMARK 500 +REMARK 500 DISTANCE CUTOFF: +REMARK 500 2.2 ANGSTROMS FOR CONTACTS NOT INVOLVING HYDROGEN ATOMS +REMARK 500 1.6 ANGSTROMS FOR CONTACTS INVOLVING HYDROGEN ATOMS +REMARK 500 +REMARK 500 ATM1 RES C SSEQI ATM2 RES C SSEQI SSYMOP DISTANCE +REMARK 500 OE2 GLU A 16 NH1 ARG A 72 1554 2.02 +REMARK 500 NZ LYS A 48 OXT GLY A 76 4467 2.16 +REMARK 500 +REMARK 500 REMARK: NULL +REMARK 500 +REMARK 500 GEOMETRY AND STEREOCHEMISTRY +REMARK 500 SUBTOPIC: COVALENT BOND ANGLES +REMARK 500 +REMARK 500 THE STEREOCHEMICAL PARAMETERS OF THE FOLLOWING RESIDUES +REMARK 500 HAVE VALUES WHICH DEVIATE FROM EXPECTED VALUES BY MORE +REMARK 500 THAN 6*RMSD (M=MODEL NUMBER; RES=RESIDUE NAME; C=CHAIN +REMARK 500 IDENTIFIER; SSEQ=SEQUENCE NUMBER; I=INSERTION CODE). +REMARK 500 +REMARK 500 STANDARD TABLE: +REMARK 500 FORMAT: (10X,I3,1X,A3,1X,A1,I4,A1,3(1X,A4,2X),12X,F5.1) +REMARK 500 +REMARK 500 EXPECTED VALUES PROTEIN: ENGH AND HUBER, 1999 +REMARK 500 EXPECTED VALUES NUCLEIC ACID: CLOWNEY ET AL 1996 +REMARK 500 +REMARK 500 M RES CSSEQI ATM1 ATM2 ATM3 +REMARK 500 LEU A 15 CA - CB - CG ANGL. DEV. = 14.0 DEGREES +REMARK 500 ARG A 54 CD - NE - CZ ANGL. DEV. = 12.4 DEGREES +REMARK 500 ARG A 54 NE - CZ - NH1 ANGL. DEV. = 5.5 DEGREES +REMARK 500 +REMARK 500 REMARK: NULL +DBREF 1UBQ A 1 76 UNP P62988 UBIQ_HUMAN 1 76 +SEQRES 1 A 76 MET GLN ILE PHE VAL LYS THR LEU THR GLY LYS THR ILE +SEQRES 2 A 76 THR LEU GLU VAL GLU PRO SER ASP THR ILE GLU ASN VAL +SEQRES 3 A 76 LYS ALA LYS ILE GLN ASP LYS GLU GLY ILE PRO PRO ASP +SEQRES 4 A 76 GLN GLN ARG LEU ILE PHE ALA GLY LYS GLN LEU GLU ASP +SEQRES 5 A 76 GLY ARG THR LEU SER ASP TYR ASN ILE GLN LYS GLU SER +SEQRES 6 A 76 THR LEU HIS LEU VAL LEU ARG LEU ARG GLY GLY +FORMUL 2 HOH *58(H2 O) +HELIX 1 H1 ILE A 23 GLU A 34 1 12 +HELIX 2 H2 LEU A 56 TYR A 59 5 4 +SHEET 1 BET 5 GLY A 10 VAL A 17 0 +SHEET 2 BET 5 MET A 1 THR A 7 -1 +SHEET 3 BET 5 GLU A 64 ARG A 72 1 +SHEET 4 BET 5 GLN A 40 PHE A 45 -1 +SHEET 5 BET 5 LYS A 48 LEU A 50 -1 +CRYST1 50.840 42.770 28.950 90.00 90.00 90.00 P 21 21 21 4 +ORIGX1 1.000000 0.000000 0.000000 0.00000 +ORIGX2 0.000000 1.000000 0.000000 0.00000 +ORIGX3 0.000000 0.000000 1.000000 0.00000 +SCALE1 0.019670 0.000000 0.000000 0.00000 +SCALE2 0.000000 0.023381 0.000000 0.00000 +SCALE3 0.000000 0.000000 0.034542 0.00000 +ATOM 1 N MET A 1 27.340 24.430 2.614 1.00 9.67 N +ATOM 2 CA MET A 1 26.266 25.413 2.842 1.00 10.38 C +ATOM 3 C MET A 1 26.913 26.639 3.531 1.00 9.62 C +ATOM 4 O MET A 1 27.886 26.463 4.263 1.00 9.62 O +ATOM 5 CB MET A 1 25.112 24.880 3.649 1.00 13.77 C +ATOM 6 CG MET A 1 25.353 24.860 5.134 1.00 16.29 C +ATOM 7 SD MET A 1 23.930 23.959 5.904 1.00 17.17 S +ATOM 8 CE MET A 1 24.447 23.984 7.620 1.00 16.11 C +ATOM 9 N GLN A 2 26.335 27.770 3.258 1.00 9.27 N +ATOM 10 CA GLN A 2 26.850 29.021 3.898 1.00 9.07 C +ATOM 11 C GLN A 2 26.100 29.253 5.202 1.00 8.72 C +ATOM 12 O GLN A 2 24.865 29.024 5.330 1.00 8.22 O +ATOM 13 CB GLN A 2 26.733 30.148 2.905 1.00 14.46 C +ATOM 14 CG GLN A 2 26.882 31.546 3.409 1.00 17.01 C +ATOM 15 CD GLN A 2 26.786 32.562 2.270 1.00 20.10 C +ATOM 16 OE1 GLN A 2 27.783 33.160 1.870 1.00 21.89 O +ATOM 17 NE2 GLN A 2 25.562 32.733 1.806 1.00 19.49 N +ATOM 18 N ILE A 3 26.849 29.656 6.217 1.00 5.87 N +ATOM 19 CA ILE A 3 26.235 30.058 7.497 1.00 5.07 C +ATOM 20 C ILE A 3 26.882 31.428 7.862 1.00 4.01 C +ATOM 21 O ILE A 3 27.906 31.711 7.264 1.00 4.61 O +ATOM 22 CB ILE A 3 26.344 29.050 8.645 1.00 6.55 C +ATOM 23 CG1 ILE A 3 27.810 28.748 8.999 1.00 4.72 C +ATOM 24 CG2 ILE A 3 25.491 27.771 8.287 1.00 5.58 C +ATOM 25 CD1 ILE A 3 27.967 28.087 10.417 1.00 10.83 C +ATOM 26 N PHE A 4 26.214 32.097 8.771 1.00 4.55 N +ATOM 27 CA PHE A 4 26.772 33.436 9.197 1.00 4.68 C +ATOM 28 C PHE A 4 27.151 33.362 10.650 1.00 5.30 C +ATOM 29 O PHE A 4 26.350 32.778 11.395 1.00 5.58 O +ATOM 30 CB PHE A 4 25.695 34.498 8.946 1.00 4.83 C +ATOM 31 CG PHE A 4 25.288 34.609 7.499 1.00 7.97 C +ATOM 32 CD1 PHE A 4 24.147 33.966 7.038 1.00 6.69 C +ATOM 33 CD2 PHE A 4 26.136 35.346 6.640 1.00 8.34 C +ATOM 34 CE1 PHE A 4 23.812 34.031 5.677 1.00 9.10 C +ATOM 35 CE2 PHE A 4 25.810 35.392 5.267 1.00 10.61 C +ATOM 36 CZ PHE A 4 24.620 34.778 4.853 1.00 8.90 C +ATOM 37 N VAL A 5 28.260 33.943 11.096 1.00 4.44 N +ATOM 38 CA VAL A 5 28.605 33.965 12.503 1.00 3.87 C +ATOM 39 C VAL A 5 28.638 35.461 12.900 1.00 4.93 C +ATOM 40 O VAL A 5 29.522 36.103 12.320 1.00 6.84 O +ATOM 41 CB VAL A 5 29.963 33.317 12.814 1.00 2.99 C +ATOM 42 CG1 VAL A 5 30.211 33.394 14.304 1.00 5.28 C +ATOM 43 CG2 VAL A 5 29.957 31.838 12.352 1.00 9.13 C +ATOM 44 N LYS A 6 27.751 35.867 13.740 1.00 6.04 N +ATOM 45 CA LYS A 6 27.691 37.315 14.143 1.00 6.12 C +ATOM 46 C LYS A 6 28.469 37.475 15.420 1.00 6.57 C +ATOM 47 O LYS A 6 28.213 36.753 16.411 1.00 5.76 O +ATOM 48 CB LYS A 6 26.219 37.684 14.307 1.00 7.45 C +ATOM 49 CG LYS A 6 25.884 39.139 14.615 1.00 11.12 C +ATOM 50 CD LYS A 6 24.348 39.296 14.642 1.00 14.54 C +ATOM 51 CE LYS A 6 23.865 40.723 14.749 1.00 18.84 C +ATOM 52 NZ LYS A 6 22.375 40.720 14.907 1.00 20.55 N +ATOM 53 N THR A 7 29.426 38.430 15.446 1.00 7.41 N +ATOM 54 CA THR A 7 30.225 38.643 16.662 1.00 7.48 C +ATOM 55 C THR A 7 29.664 39.839 17.434 1.00 8.75 C +ATOM 56 O THR A 7 28.850 40.565 16.859 1.00 8.58 O +ATOM 57 CB THR A 7 31.744 38.879 16.299 1.00 9.61 C +ATOM 58 OG1 THR A 7 31.737 40.257 15.824 1.00 11.78 O +ATOM 59 CG2 THR A 7 32.260 37.969 15.171 1.00 9.17 C +ATOM 60 N LEU A 8 30.132 40.069 18.642 1.00 9.84 N +ATOM 61 CA LEU A 8 29.607 41.180 19.467 1.00 14.15 C +ATOM 62 C LEU A 8 30.075 42.538 18.984 1.00 17.37 C +ATOM 63 O LEU A 8 29.586 43.570 19.483 1.00 17.01 O +ATOM 64 CB LEU A 8 29.919 40.890 20.938 1.00 16.63 C +ATOM 65 CG LEU A 8 29.183 39.722 21.581 1.00 18.88 C +ATOM 66 CD1 LEU A 8 29.308 39.750 23.095 1.00 19.31 C +ATOM 67 CD2 LEU A 8 27.700 39.721 21.228 1.00 18.59 C +ATOM 68 N THR A 9 30.991 42.571 17.998 1.00 18.33 N +ATOM 69 CA THR A 9 31.422 43.940 17.553 1.00 19.24 C +ATOM 70 C THR A 9 30.755 44.351 16.277 1.00 19.48 C +ATOM 71 O THR A 9 31.207 45.268 15.566 1.00 23.14 O +ATOM 72 CB THR A 9 32.979 43.918 17.445 1.00 18.97 C +ATOM 73 OG1 THR A 9 33.174 43.067 16.265 1.00 20.24 O +ATOM 74 CG2 THR A 9 33.657 43.319 18.672 1.00 19.70 C +ATOM 75 N GLY A 10 29.721 43.673 15.885 1.00 19.43 N +ATOM 76 CA GLY A 10 28.978 43.960 14.678 1.00 18.74 C +ATOM 77 C GLY A 10 29.604 43.507 13.393 1.00 17.62 C +ATOM 78 O GLY A 10 29.219 43.981 12.301 1.00 19.74 O +ATOM 79 N LYS A 11 30.563 42.623 13.495 1.00 13.56 N +ATOM 80 CA LYS A 11 31.191 42.012 12.331 1.00 11.91 C +ATOM 81 C LYS A 11 30.459 40.666 12.130 1.00 10.18 C +ATOM 82 O LYS A 11 30.253 39.991 13.133 1.00 9.10 O +ATOM 83 CB LYS A 11 32.672 41.717 12.505 1.00 13.43 C +ATOM 84 CG LYS A 11 33.280 41.086 11.227 1.00 16.69 C +ATOM 85 CD LYS A 11 34.762 40.799 11.470 1.00 17.92 C +ATOM 86 CE LYS A 11 35.614 40.847 10.240 1.00 20.81 C +ATOM 87 NZ LYS A 11 35.100 40.073 9.101 1.00 21.93 N +ATOM 88 N THR A 12 30.163 40.338 10.886 1.00 9.63 N +ATOM 89 CA THR A 12 29.542 39.020 10.653 1.00 9.85 C +ATOM 90 C THR A 12 30.494 38.261 9.729 1.00 11.66 C +ATOM 91 O THR A 12 30.849 38.850 8.706 1.00 12.33 O +ATOM 92 CB THR A 12 28.113 39.049 10.015 1.00 10.85 C +ATOM 93 OG1 THR A 12 27.280 39.722 10.996 1.00 10.91 O +ATOM 94 CG2 THR A 12 27.588 37.635 9.715 1.00 9.63 C +ATOM 95 N ILE A 13 30.795 37.015 10.095 1.00 10.42 N +ATOM 96 CA ILE A 13 31.720 36.289 9.176 1.00 11.84 C +ATOM 97 C ILE A 13 30.955 35.211 8.459 1.00 10.55 C +ATOM 98 O ILE A 13 30.025 34.618 9.040 1.00 11.92 O +ATOM 99 CB ILE A 13 32.995 35.883 9.934 1.00 14.86 C +ATOM 100 CG1 ILE A 13 33.306 34.381 9.840 1.00 14.87 C +ATOM 101 CG2 ILE A 13 33.109 36.381 11.435 1.00 17.08 C +ATOM 102 CD1 ILE A 13 34.535 34.028 10.720 1.00 16.46 C +ATOM 103 N THR A 14 31.244 34.986 7.197 1.00 9.39 N +ATOM 104 CA THR A 14 30.505 33.884 6.512 1.00 9.63 C +ATOM 105 C THR A 14 31.409 32.680 6.446 1.00 11.20 C +ATOM 106 O THR A 14 32.619 32.812 6.125 1.00 11.63 O +ATOM 107 CB THR A 14 30.091 34.393 5.078 1.00 10.38 C +ATOM 108 OG1 THR A 14 31.440 34.513 4.487 1.00 16.30 O +ATOM 109 CG2 THR A 14 29.420 35.756 5.119 1.00 11.66 C +ATOM 110 N LEU A 15 30.884 31.485 6.666 1.00 8.29 N +ATOM 111 CA LEU A 15 31.677 30.275 6.639 1.00 9.03 C +ATOM 112 C LEU A 15 31.022 29.288 5.665 1.00 8.59 C +ATOM 113 O LEU A 15 29.809 29.395 5.545 1.00 7.79 O +ATOM 114 CB LEU A 15 31.562 29.686 8.045 1.00 11.08 C +ATOM 115 CG LEU A 15 32.631 29.444 9.060 1.00 15.79 C +ATOM 116 CD1 LEU A 15 33.814 30.390 9.030 1.00 15.88 C +ATOM 117 CD2 LEU A 15 31.945 29.449 10.436 1.00 15.27 C +ATOM 118 N GLU A 16 31.834 28.412 5.125 1.00 11.04 N +ATOM 119 CA GLU A 16 31.220 27.341 4.275 1.00 11.50 C +ATOM 120 C GLU A 16 31.440 26.079 5.080 1.00 10.13 C +ATOM 121 O GLU A 16 32.576 25.802 5.461 1.00 9.83 O +ATOM 122 CB GLU A 16 31.827 27.262 2.894 1.00 17.22 C +ATOM 123 CG GLU A 16 31.363 28.410 1.962 1.00 23.33 C +ATOM 124 CD GLU A 16 31.671 28.291 0.498 1.00 26.99 C +ATOM 125 OE1 GLU A 16 30.869 28.621 -0.366 1.00 28.86 O +ATOM 126 OE2 GLU A 16 32.835 27.861 0.278 1.00 28.90 O +ATOM 127 N VAL A 17 30.310 25.458 5.384 1.00 8.99 N +ATOM 128 CA VAL A 17 30.288 24.245 6.193 1.00 8.85 C +ATOM 129 C VAL A 17 29.279 23.227 5.641 1.00 8.04 C +ATOM 130 O VAL A 17 28.478 23.522 4.725 1.00 8.99 O +ATOM 131 CB VAL A 17 29.903 24.590 7.665 1.00 9.78 C +ATOM 132 CG1 VAL A 17 30.862 25.496 8.389 1.00 12.05 C +ATOM 133 CG2 VAL A 17 28.476 25.135 7.705 1.00 10.54 C +ATOM 134 N GLU A 18 29.380 22.057 6.232 1.00 7.29 N +ATOM 135 CA GLU A 18 28.468 20.940 5.980 1.00 7.08 C +ATOM 136 C GLU A 18 27.819 20.609 7.316 1.00 6.45 C +ATOM 137 O GLU A 18 28.449 20.674 8.360 1.00 5.28 O +ATOM 138 CB GLU A 18 29.213 19.697 5.506 1.00 10.28 C +ATOM 139 CG GLU A 18 29.728 19.755 4.060 1.00 12.65 C +ATOM 140 CD GLU A 18 28.754 20.061 2.978 1.00 14.15 C +ATOM 141 OE1 GLU A 18 27.546 19.992 2.985 1.00 14.33 O +ATOM 142 OE2 GLU A 18 29.336 20.423 1.904 1.00 18.17 O +ATOM 143 N PRO A 19 26.559 20.220 7.288 1.00 7.24 N +ATOM 144 CA PRO A 19 25.829 19.825 8.494 1.00 7.07 C +ATOM 145 C PRO A 19 26.541 18.732 9.251 1.00 6.65 C +ATOM 146 O PRO A 19 26.333 18.536 10.457 1.00 6.37 O +ATOM 147 CB PRO A 19 24.469 19.332 7.952 1.00 7.61 C +ATOM 148 CG PRO A 19 24.299 20.134 6.704 1.00 8.16 C +ATOM 149 CD PRO A 19 25.714 20.108 6.073 1.00 7.49 C +ATOM 150 N SER A 20 27.361 17.959 8.559 1.00 6.80 N +ATOM 151 CA SER A 20 28.054 16.835 9.210 1.00 6.28 C +ATOM 152 C SER A 20 29.258 17.318 9.984 1.00 8.45 C +ATOM 153 O SER A 20 29.930 16.477 10.606 1.00 7.26 O +ATOM 154 CB SER A 20 28.523 15.820 8.182 1.00 8.57 C +ATOM 155 OG SER A 20 28.946 16.445 6.967 1.00 11.13 O +ATOM 156 N ASP A 21 29.599 18.599 9.828 1.00 7.50 N +ATOM 157 CA ASP A 21 30.796 19.083 10.566 1.00 7.70 C +ATOM 158 C ASP A 21 30.491 19.162 12.040 1.00 7.08 C +ATOM 159 O ASP A 21 29.367 19.523 12.441 1.00 8.11 O +ATOM 160 CB ASP A 21 31.155 20.515 10.048 1.00 11.00 C +ATOM 161 CG ASP A 21 31.923 20.436 8.755 1.00 15.32 C +ATOM 162 OD1 ASP A 21 32.493 19.374 8.456 1.00 18.03 O +ATOM 163 OD2 ASP A 21 31.838 21.402 7.968 1.00 14.36 O +ATOM 164 N THR A 22 31.510 18.936 12.852 1.00 5.37 N +ATOM 165 CA THR A 22 31.398 19.064 14.286 1.00 6.01 C +ATOM 166 C THR A 22 31.593 20.553 14.655 1.00 8.01 C +ATOM 167 O THR A 22 32.159 21.311 13.861 1.00 8.11 O +ATOM 168 CB THR A 22 32.492 18.193 14.995 1.00 8.92 C +ATOM 169 OG1 THR A 22 33.778 18.739 14.516 1.00 10.22 O +ATOM 170 CG2 THR A 22 32.352 16.700 14.630 1.00 9.65 C +ATOM 171 N ILE A 23 31.113 20.863 15.860 1.00 8.32 N +ATOM 172 CA ILE A 23 31.288 22.201 16.417 1.00 9.92 C +ATOM 173 C ILE A 23 32.776 22.519 16.577 1.00 10.01 C +ATOM 174 O ILE A 23 33.233 23.659 16.384 1.00 8.71 O +ATOM 175 CB ILE A 23 30.520 22.300 17.764 1.00 10.78 C +ATOM 176 CG1 ILE A 23 29.006 22.043 17.442 1.00 11.38 C +ATOM 177 CG2 ILE A 23 30.832 23.699 18.358 1.00 10.90 C +ATOM 178 CD1 ILE A 23 28.407 22.948 16.366 1.00 12.30 C +ATOM 179 N GLU A 24 33.548 21.526 16.950 1.00 9.54 N +ATOM 180 CA GLU A 24 35.031 21.722 17.069 1.00 11.81 C +ATOM 181 C GLU A 24 35.615 22.190 15.759 1.00 11.14 C +ATOM 182 O GLU A 24 36.532 23.046 15.724 1.00 10.62 O +ATOM 183 CB GLU A 24 35.667 20.383 17.447 1.00 19.24 C +ATOM 184 CG GLU A 24 37.128 20.293 17.872 1.00 27.76 C +ATOM 185 CD GLU A 24 37.561 18.851 18.082 1.00 32.92 C +ATOM 186 OE1 GLU A 24 37.758 18.024 17.195 1.00 34.80 O +ATOM 187 OE2 GLU A 24 37.628 18.599 19.313 1.00 36.51 O +ATOM 188 N ASN A 25 35.139 21.624 14.662 1.00 9.43 N +ATOM 189 CA ASN A 25 35.590 21.945 13.302 1.00 10.96 C +ATOM 190 C ASN A 25 35.238 23.382 12.920 1.00 9.68 C +ATOM 191 O ASN A 25 36.066 24.109 12.333 1.00 9.33 O +ATOM 192 CB ASN A 25 35.064 20.957 12.255 1.00 16.78 C +ATOM 193 CG ASN A 25 35.541 21.418 10.871 1.00 22.31 C +ATOM 194 OD1 ASN A 25 36.772 21.623 10.676 1.00 25.66 O +ATOM 195 ND2 ASN A 25 34.628 21.595 9.920 1.00 24.70 N +ATOM 196 N VAL A 26 34.007 23.745 13.250 1.00 6.52 N +ATOM 197 CA VAL A 26 33.533 25.097 12.978 1.00 5.53 C +ATOM 198 C VAL A 26 34.441 26.099 13.684 1.00 4.42 C +ATOM 199 O VAL A 26 34.883 27.090 13.093 1.00 3.40 O +ATOM 200 CB VAL A 26 32.060 25.257 13.364 1.00 3.86 C +ATOM 201 CG1 VAL A 26 31.684 26.749 13.342 1.00 7.25 C +ATOM 202 CG2 VAL A 26 31.152 24.421 12.477 1.00 8.12 C +ATOM 203 N LYS A 27 34.734 25.822 14.949 1.00 2.64 N +ATOM 204 CA LYS A 27 35.596 26.715 15.736 1.00 4.14 C +ATOM 205 C LYS A 27 36.975 26.826 15.107 1.00 5.58 C +ATOM 206 O LYS A 27 37.579 27.926 15.159 1.00 4.11 O +ATOM 207 CB LYS A 27 35.715 26.203 17.172 1.00 3.97 C +ATOM 208 CG LYS A 27 34.343 26.445 17.898 1.00 7.45 C +ATOM 209 CD LYS A 27 34.509 26.077 19.360 1.00 9.02 C +ATOM 210 CE LYS A 27 33.206 26.311 20.122 1.00 12.90 C +ATOM 211 NZ LYS A 27 33.455 25.910 21.546 1.00 15.47 N +ATOM 212 N ALA A 28 37.499 25.743 14.571 1.00 6.61 N +ATOM 213 CA ALA A 28 38.794 25.761 13.880 1.00 7.74 C +ATOM 214 C ALA A 28 38.728 26.591 12.611 1.00 9.17 C +ATOM 215 O ALA A 28 39.704 27.346 12.277 1.00 11.45 O +ATOM 216 CB ALA A 28 39.285 24.336 13.566 1.00 7.68 C +ATOM 217 N LYS A 29 37.633 26.543 11.867 1.00 8.96 N +ATOM 218 CA LYS A 29 37.471 27.391 10.668 1.00 7.90 C +ATOM 219 C LYS A 29 37.441 28.882 11.052 1.00 6.92 C +ATOM 220 O LYS A 29 38.020 29.772 10.382 1.00 6.87 O +ATOM 221 CB LYS A 29 36.193 27.058 9.911 1.00 10.28 C +ATOM 222 CG LYS A 29 36.153 25.620 9.409 1.00 14.94 C +ATOM 223 CD LYS A 29 34.758 25.280 8.900 1.00 19.69 C +ATOM 224 CE LYS A 29 34.793 24.264 7.767 1.00 22.63 C +ATOM 225 NZ LYS A 29 34.914 24.944 6.441 1.00 24.98 N +ATOM 226 N ILE A 30 36.811 29.170 12.192 1.00 4.57 N +ATOM 227 CA ILE A 30 36.731 30.570 12.645 1.00 5.58 C +ATOM 228 C ILE A 30 38.148 30.981 13.069 1.00 7.26 C +ATOM 229 O ILE A 30 38.544 32.150 12.856 1.00 9.46 O +ATOM 230 CB ILE A 30 35.708 30.776 13.806 1.00 5.36 C +ATOM 231 CG1 ILE A 30 34.228 30.630 13.319 1.00 2.94 C +ATOM 232 CG2 ILE A 30 35.874 32.138 14.512 1.00 2.78 C +ATOM 233 CD1 ILE A 30 33.284 30.504 14.552 1.00 2.00 C +ATOM 234 N GLN A 31 38.883 30.110 13.713 1.00 7.06 N +ATOM 235 CA GLN A 31 40.269 30.508 14.115 1.00 8.67 C +ATOM 236 C GLN A 31 41.092 30.808 12.851 1.00 10.90 C +ATOM 237 O GLN A 31 41.828 31.808 12.681 1.00 9.63 O +ATOM 238 CB GLN A 31 40.996 29.399 14.865 1.00 9.12 C +ATOM 239 CG GLN A 31 42.445 29.848 15.182 1.00 10.76 C +ATOM 240 CD GLN A 31 43.090 28.828 16.095 1.00 13.78 C +ATOM 241 OE1 GLN A 31 42.770 27.655 15.906 1.00 14.48 O +ATOM 242 NE2 GLN A 31 43.898 29.252 17.050 1.00 14.76 N +ATOM 243 N ASP A 32 41.001 29.878 11.931 1.00 10.93 N +ATOM 244 CA ASP A 32 41.718 30.022 10.643 1.00 14.01 C +ATOM 245 C ASP A 32 41.399 31.338 9.967 1.00 14.04 C +ATOM 246 O ASP A 32 42.260 32.036 9.381 1.00 13.39 O +ATOM 247 CB ASP A 32 41.398 28.780 9.810 1.00 18.01 C +ATOM 248 CG ASP A 32 42.626 28.557 8.928 1.00 24.33 C +ATOM 249 OD1 ASP A 32 43.666 28.262 9.539 1.00 26.29 O +ATOM 250 OD2 ASP A 32 42.430 28.812 7.728 1.00 25.17 O +ATOM 251 N LYS A 33 40.117 31.750 9.988 1.00 14.22 N +ATOM 252 CA LYS A 33 39.808 32.994 9.233 1.00 14.00 C +ATOM 253 C LYS A 33 39.837 34.271 9.995 1.00 12.37 C +ATOM 254 O LYS A 33 40.164 35.323 9.345 1.00 12.17 O +ATOM 255 CB LYS A 33 38.615 32.801 8.320 1.00 18.62 C +ATOM 256 CG LYS A 33 37.220 32.822 8.827 1.00 24.00 C +ATOM 257 CD LYS A 33 36.351 33.613 7.838 1.00 27.61 C +ATOM 258 CE LYS A 33 36.322 32.944 6.477 1.00 27.64 C +ATOM 259 NZ LYS A 33 35.768 33.945 5.489 1.00 30.06 N +ATOM 260 N GLU A 34 39.655 34.335 11.285 1.00 10.11 N +ATOM 261 CA GLU A 34 39.676 35.547 12.072 1.00 10.07 C +ATOM 262 C GLU A 34 40.675 35.527 13.200 1.00 9.32 C +ATOM 263 O GLU A 34 40.814 36.528 13.911 1.00 11.61 O +ATOM 264 CB GLU A 34 38.290 35.814 12.698 1.00 14.77 C +ATOM 265 CG GLU A 34 37.156 35.985 11.688 1.00 18.75 C +ATOM 266 CD GLU A 34 37.192 37.361 11.033 1.00 22.28 C +ATOM 267 OE1 GLU A 34 37.519 38.360 11.645 1.00 21.95 O +ATOM 268 OE2 GLU A 34 36.861 37.320 9.822 1.00 25.19 O +ATOM 269 N GLY A 35 41.317 34.393 13.432 1.00 7.22 N +ATOM 270 CA GLY A 35 42.345 34.269 14.431 1.00 6.29 C +ATOM 271 C GLY A 35 41.949 34.076 15.842 1.00 6.93 C +ATOM 272 O GLY A 35 42.829 34.000 16.739 1.00 7.41 O +ATOM 273 N ILE A 36 40.642 33.916 16.112 1.00 5.86 N +ATOM 274 CA ILE A 36 40.226 33.716 17.509 1.00 6.07 C +ATOM 275 C ILE A 36 40.449 32.278 17.945 1.00 6.36 C +ATOM 276 O ILE A 36 39.936 31.336 17.315 1.00 6.18 O +ATOM 277 CB ILE A 36 38.693 34.106 17.595 1.00 7.47 C +ATOM 278 CG1 ILE A 36 38.471 35.546 17.045 1.00 8.52 C +ATOM 279 CG2 ILE A 36 38.146 33.932 19.027 1.00 7.36 C +ATOM 280 CD1 ILE A 36 36.958 35.746 16.680 1.00 9.49 C +ATOM 281 N PRO A 37 41.189 32.085 19.031 1.00 8.65 N +ATOM 282 CA PRO A 37 41.461 30.751 19.594 1.00 9.18 C +ATOM 283 C PRO A 37 40.168 30.026 19.918 1.00 9.85 C +ATOM 284 O PRO A 37 39.264 30.662 20.521 1.00 8.51 O +ATOM 285 CB PRO A 37 42.195 31.142 20.913 1.00 11.42 C +ATOM 286 CG PRO A 37 42.904 32.414 20.553 1.00 9.27 C +ATOM 287 CD PRO A 37 41.822 33.188 19.813 1.00 8.33 C +ATOM 288 N PRO A 38 40.059 28.758 19.607 1.00 8.71 N +ATOM 289 CA PRO A 38 38.817 28.020 19.889 1.00 9.08 C +ATOM 290 C PRO A 38 38.421 28.048 21.341 1.00 9.28 C +ATOM 291 O PRO A 38 37.213 28.036 21.704 1.00 6.50 O +ATOM 292 CB PRO A 38 39.090 26.629 19.325 1.00 10.31 C +ATOM 293 CG PRO A 38 40.082 26.904 18.198 1.00 10.81 C +ATOM 294 CD PRO A 38 41.035 27.909 18.879 1.00 12.00 C +ATOM 295 N ASP A 39 39.374 28.090 22.240 1.00 11.20 N +ATOM 296 CA ASP A 39 39.063 28.063 23.695 1.00 14.96 C +ATOM 297 C ASP A 39 38.365 29.335 24.159 1.00 13.99 C +ATOM 298 O ASP A 39 37.684 29.390 25.221 1.00 13.75 O +ATOM 299 CB ASP A 39 40.340 27.692 24.468 1.00 24.16 C +ATOM 300 CG ASP A 39 40.559 28.585 25.675 1.00 31.06 C +ATOM 301 OD1 ASP A 39 40.716 29.809 25.456 1.00 35.55 O +ATOM 302 OD2 ASP A 39 40.549 28.090 26.840 1.00 34.22 O +ATOM 303 N GLN A 40 38.419 30.373 23.341 1.00 11.60 N +ATOM 304 CA GLN A 40 37.738 31.637 23.712 1.00 10.76 C +ATOM 305 C GLN A 40 36.334 31.742 23.087 1.00 8.01 C +ATOM 306 O GLN A 40 35.574 32.618 23.483 1.00 8.96 O +ATOM 307 CB GLN A 40 38.528 32.854 23.182 1.00 11.14 C +ATOM 308 CG GLN A 40 39.919 32.854 23.840 1.00 14.85 C +ATOM 309 CD GLN A 40 40.760 34.036 23.394 1.00 16.11 C +ATOM 310 OE1 GLN A 40 41.975 34.008 23.624 1.00 20.52 O +ATOM 311 NE2 GLN A 40 40.140 35.007 22.775 1.00 18.16 N +ATOM 312 N GLN A 41 36.000 30.860 22.172 1.00 6.52 N +ATOM 313 CA GLN A 41 34.738 30.875 21.473 1.00 3.87 C +ATOM 314 C GLN A 41 33.589 30.189 22.181 1.00 4.79 C +ATOM 315 O GLN A 41 33.580 29.009 22.499 1.00 6.34 O +ATOM 316 CB GLN A 41 34.876 30.237 20.066 1.00 4.20 C +ATOM 317 CG GLN A 41 36.012 30.860 19.221 1.00 3.20 C +ATOM 318 CD GLN A 41 36.083 30.194 17.875 1.00 4.89 C +ATOM 319 OE1 GLN A 41 35.048 29.702 17.393 1.00 5.21 O +ATOM 320 NE2 GLN A 41 37.228 30.126 17.233 1.00 7.13 N +ATOM 321 N ARG A 42 32.478 30.917 22.269 1.00 5.73 N +ATOM 322 CA ARG A 42 31.200 30.329 22.780 1.00 6.97 C +ATOM 323 C ARG A 42 30.210 30.509 21.650 1.00 7.15 C +ATOM 324 O ARG A 42 29.978 31.726 21.269 1.00 7.33 O +ATOM 325 CB ARG A 42 30.847 30.931 24.118 1.00 13.23 C +ATOM 326 CG ARG A 42 29.412 30.796 24.598 1.00 21.27 C +ATOM 327 CD ARG A 42 29.271 31.314 26.016 1.00 26.14 C +ATOM 328 NE ARG A 42 27.875 31.317 26.443 1.00 32.26 N +ATOM 329 CZ ARG A 42 27.132 32.423 26.574 1.00 34.32 C +ATOM 330 NH1 ARG A 42 27.630 33.656 26.461 1.00 35.30 N +ATOM 331 NH2 ARG A 42 25.810 32.299 26.732 1.00 36.39 N +ATOM 332 N LEU A 43 29.694 29.436 21.054 1.00 4.65 N +ATOM 333 CA LEU A 43 28.762 29.573 19.906 1.00 3.51 C +ATOM 334 C LEU A 43 27.331 29.317 20.364 1.00 5.56 C +ATOM 335 O LEU A 43 27.101 28.346 21.097 1.00 4.19 O +ATOM 336 CB LEU A 43 29.151 28.655 18.755 1.00 3.74 C +ATOM 337 CG LEU A 43 30.416 28.912 17.980 1.00 6.32 C +ATOM 338 CD1 LEU A 43 30.738 27.693 17.122 1.00 9.55 C +ATOM 339 CD2 LEU A 43 30.205 30.168 17.129 1.00 6.41 C +ATOM 340 N ILE A 44 26.436 30.232 20.004 1.00 4.58 N +ATOM 341 CA ILE A 44 25.034 30.170 20.401 1.00 5.55 C +ATOM 342 C ILE A 44 24.101 30.149 19.196 1.00 5.46 C +ATOM 343 O ILE A 44 24.196 30.948 18.287 1.00 6.04 O +ATOM 344 CB ILE A 44 24.639 31.426 21.286 1.00 6.80 C +ATOM 345 CG1 ILE A 44 25.646 31.670 22.421 1.00 10.31 C +ATOM 346 CG2 ILE A 44 23.181 31.309 21.824 1.00 7.39 C +ATOM 347 CD1 ILE A 44 25.778 30.436 23.356 1.00 13.90 C +ATOM 348 N PHE A 45 23.141 29.187 19.241 1.00 6.75 N +ATOM 349 CA PHE A 45 22.126 29.062 18.183 1.00 4.70 C +ATOM 350 C PHE A 45 20.835 28.629 18.904 1.00 6.34 C +ATOM 351 O PHE A 45 20.821 27.734 19.749 1.00 5.45 O +ATOM 352 CB PHE A 45 22.494 28.057 17.109 1.00 5.51 C +ATOM 353 CG PHE A 45 21.447 27.869 16.026 1.00 5.98 C +ATOM 354 CD1 PHE A 45 21.325 28.813 15.005 1.00 6.86 C +ATOM 355 CD2 PHE A 45 20.638 26.735 16.053 1.00 5.87 C +ATOM 356 CE1 PHE A 45 20.369 28.648 14.001 1.00 6.68 C +ATOM 357 CE2 PHE A 45 19.677 26.539 15.051 1.00 6.64 C +ATOM 358 CZ PHE A 45 19.593 27.465 14.021 1.00 6.84 C +ATOM 359 N ALA A 46 19.810 29.378 18.578 1.00 6.53 N +ATOM 360 CA ALA A 46 18.443 29.143 19.083 1.00 7.15 C +ATOM 361 C ALA A 46 18.453 28.941 20.591 1.00 9.00 C +ATOM 362 O ALA A 46 17.860 27.994 21.128 1.00 11.15 O +ATOM 363 CB ALA A 46 17.864 27.977 18.346 1.00 8.99 C +ATOM 364 N GLY A 47 19.172 29.808 21.243 1.00 9.35 N +ATOM 365 CA GLY A 47 19.399 29.894 22.655 1.00 11.68 C +ATOM 366 C GLY A 47 20.083 28.729 23.321 1.00 11.14 C +ATOM 367 O GLY A 47 19.991 28.584 24.561 1.00 13.93 O +ATOM 368 N LYS A 48 20.801 27.931 22.578 1.00 10.47 N +ATOM 369 CA LYS A 48 21.550 26.796 23.133 1.00 8.82 C +ATOM 370 C LYS A 48 23.046 27.087 22.913 1.00 7.68 C +ATOM 371 O LYS A 48 23.383 27.627 21.870 1.00 6.47 O +ATOM 372 CB LYS A 48 21.242 25.519 22.391 1.00 9.74 C +ATOM 373 CG LYS A 48 19.762 25.077 22.455 1.00 14.14 C +ATOM 374 CD LYS A 48 19.634 23.885 21.531 1.00 16.32 C +ATOM 375 CE LYS A 48 18.791 24.221 20.313 1.00 20.04 C +ATOM 376 NZ LYS A 48 17.440 24.655 20.827 1.00 23.92 N +ATOM 377 N GLN A 49 23.880 26.727 23.851 1.00 8.89 N +ATOM 378 CA GLN A 49 25.349 26.872 23.643 1.00 7.18 C +ATOM 379 C GLN A 49 25.743 25.586 22.922 1.00 8.23 C +ATOM 380 O GLN A 49 25.325 24.489 23.378 1.00 9.70 O +ATOM 381 CB GLN A 49 26.070 27.025 24.960 1.00 11.67 C +ATOM 382 CG GLN A 49 27.553 27.356 24.695 1.00 15.82 C +ATOM 383 CD GLN A 49 28.262 27.576 26.020 1.00 20.21 C +ATOM 384 OE1 GLN A 49 29.189 26.840 26.335 1.00 23.23 O +ATOM 385 NE2 GLN A 49 27.777 28.585 26.739 1.00 20.67 N +ATOM 386 N LEU A 50 26.465 25.689 21.833 1.00 6.51 N +ATOM 387 CA LEU A 50 26.826 24.521 21.012 1.00 7.41 C +ATOM 388 C LEU A 50 27.994 23.781 21.643 1.00 8.27 C +ATOM 389 O LEU A 50 28.904 24.444 22.098 1.00 8.34 O +ATOM 390 CB LEU A 50 27.043 24.992 19.571 1.00 7.13 C +ATOM 391 CG LEU A 50 25.931 25.844 18.959 1.00 7.53 C +ATOM 392 CD1 LEU A 50 26.203 26.083 17.471 1.00 8.14 C +ATOM 393 CD2 LEU A 50 24.577 25.190 19.079 1.00 9.11 C +ATOM 394 N GLU A 51 27.942 22.448 21.648 1.00 9.43 N +ATOM 395 CA GLU A 51 29.015 21.657 22.288 1.00 11.90 C +ATOM 396 C GLU A 51 29.942 21.106 21.240 1.00 11.49 C +ATOM 397 O GLU A 51 29.470 20.677 20.190 1.00 9.88 O +ATOM 398 CB GLU A 51 28.348 20.540 23.066 1.00 16.56 C +ATOM 399 CG GLU A 51 29.247 19.456 23.705 1.00 26.06 C +ATOM 400 CD GLU A 51 28.722 19.047 25.066 1.00 29.86 C +ATOM 401 OE1 GLU A 51 29.139 18.132 25.746 1.00 32.13 O +ATOM 402 OE2 GLU A 51 27.777 19.842 25.367 1.00 33.44 O +ATOM 403 N ASP A 52 31.233 21.090 21.459 1.00 12.71 N +ATOM 404 CA ASP A 52 32.262 20.670 20.514 1.00 16.56 C +ATOM 405 C ASP A 52 32.128 19.364 19.750 1.00 15.83 C +ATOM 406 O ASP A 52 32.546 19.317 18.558 1.00 17.21 O +ATOM 407 CB ASP A 52 33.638 20.716 21.242 1.00 21.05 C +ATOM 408 CG ASP A 52 34.174 22.129 21.354 1.00 25.12 C +ATOM 409 OD1 ASP A 52 35.252 22.322 21.958 1.00 28.37 O +ATOM 410 OD2 ASP A 52 33.544 23.086 20.883 1.00 25.82 O +ATOM 411 N GLY A 53 31.697 18.311 20.406 1.00 15.00 N +ATOM 412 CA GLY A 53 31.568 16.962 19.825 1.00 11.77 C +ATOM 413 C GLY A 53 30.320 16.698 19.051 1.00 11.10 C +ATOM 414 O GLY A 53 30.198 15.657 18.366 1.00 11.25 O +ATOM 415 N ARG A 54 29.340 17.594 19.076 1.00 8.53 N +ATOM 416 CA ARG A 54 28.108 17.439 18.276 1.00 9.05 C +ATOM 417 C ARG A 54 28.375 17.999 16.887 1.00 8.96 C +ATOM 418 O ARG A 54 29.326 18.786 16.690 1.00 11.60 O +ATOM 419 CB ARG A 54 26.926 18.191 18.892 1.00 7.97 C +ATOM 420 CG ARG A 54 26.621 17.799 20.352 1.00 9.62 C +ATOM 421 CD ARG A 54 26.010 16.370 20.280 1.00 12.20 C +ATOM 422 NE ARG A 54 26.975 15.521 20.942 1.00 18.23 N +ATOM 423 CZ ARG A 54 27.603 14.423 20.655 1.00 22.08 C +ATOM 424 NH1 ARG A 54 27.479 13.733 19.537 1.00 23.38 N +ATOM 425 NH2 ARG A 54 28.519 13.967 21.550 1.00 25.50 N +ATOM 426 N THR A 55 27.510 17.689 15.954 1.00 9.05 N +ATOM 427 CA THR A 55 27.574 18.192 14.563 1.00 9.03 C +ATOM 428 C THR A 55 26.482 19.280 14.432 1.00 8.15 C +ATOM 429 O THR A 55 25.609 19.388 15.287 1.00 5.91 O +ATOM 430 CB THR A 55 27.299 17.055 13.533 1.00 11.15 C +ATOM 431 OG1 THR A 55 25.925 16.611 13.913 1.00 11.95 O +ATOM 432 CG2 THR A 55 28.236 15.864 13.558 1.00 11.71 C +ATOM 433 N LEU A 56 26.585 20.063 13.378 1.00 6.91 N +ATOM 434 CA LEU A 56 25.594 21.109 13.072 1.00 8.29 C +ATOM 435 C LEU A 56 24.241 20.436 12.857 1.00 8.05 C +ATOM 436 O LEU A 56 23.264 20.951 13.329 1.00 10.17 O +ATOM 437 CB LEU A 56 26.084 21.888 11.833 1.00 6.60 C +ATOM 438 CG LEU A 56 27.426 22.616 11.902 1.00 7.73 C +ATOM 439 CD1 LEU A 56 27.718 23.341 10.578 1.00 9.85 C +ATOM 440 CD2 LEU A 56 27.380 23.721 12.955 1.00 8.64 C +ATOM 441 N SER A 57 24.240 19.233 12.246 1.00 8.92 N +ATOM 442 CA SER A 57 22.924 18.583 12.025 1.00 9.00 C +ATOM 443 C SER A 57 22.229 18.244 13.325 1.00 9.44 C +ATOM 444 O SER A 57 20.963 18.253 13.395 1.00 10.91 O +ATOM 445 CB SER A 57 23.059 17.326 11.154 1.00 10.32 C +ATOM 446 OG SER A 57 23.914 16.395 11.755 1.00 13.59 O +ATOM 447 N ASP A 58 22.997 17.978 14.366 1.00 9.11 N +ATOM 448 CA ASP A 58 22.418 17.638 15.693 1.00 7.91 C +ATOM 449 C ASP A 58 21.460 18.737 16.163 1.00 9.12 C +ATOM 450 O ASP A 58 20.497 18.506 16.900 1.00 8.61 O +ATOM 451 CB ASP A 58 23.461 17.331 16.741 1.00 8.41 C +ATOM 452 CG ASP A 58 24.184 16.016 16.619 1.00 11.50 C +ATOM 453 OD1 ASP A 58 25.303 15.894 17.152 1.00 10.05 O +ATOM 454 OD2 ASP A 58 23.572 15.107 15.975 1.00 11.70 O +ATOM 455 N TYR A 59 21.846 19.954 15.905 1.00 7.97 N +ATOM 456 CA TYR A 59 21.079 21.149 16.251 1.00 8.45 C +ATOM 457 C TYR A 59 20.142 21.590 15.149 1.00 10.98 C +ATOM 458 O TYR A 59 19.499 22.645 15.321 1.00 12.95 O +ATOM 459 CB TYR A 59 22.085 22.254 16.581 1.00 7.94 C +ATOM 460 CG TYR A 59 22.945 21.951 17.785 1.00 6.91 C +ATOM 461 CD1 TYR A 59 24.272 21.544 17.644 1.00 4.59 C +ATOM 462 CD2 TYR A 59 22.437 22.157 19.065 1.00 6.98 C +ATOM 463 CE1 TYR A 59 25.052 21.285 18.776 1.00 5.39 C +ATOM 464 CE2 TYR A 59 23.204 21.907 20.192 1.00 6.52 C +ATOM 465 CZ TYR A 59 24.517 21.470 20.030 1.00 6.76 C +ATOM 466 OH TYR A 59 25.248 21.302 21.191 1.00 7.63 O +ATOM 467 N ASN A 60 19.993 20.884 14.049 1.00 12.38 N +ATOM 468 CA ASN A 60 19.065 21.352 12.999 1.00 13.94 C +ATOM 469 C ASN A 60 19.442 22.745 12.510 1.00 14.16 C +ATOM 470 O ASN A 60 18.571 23.610 12.289 1.00 14.26 O +ATOM 471 CB ASN A 60 17.586 21.282 13.461 1.00 19.23 C +ATOM 472 CG ASN A 60 16.576 21.258 12.315 1.00 22.65 C +ATOM 473 OD1 ASN A 60 15.440 21.819 12.378 1.00 25.45 O +ATOM 474 ND2 ASN A 60 16.924 20.586 11.216 1.00 24.09 N +ATOM 475 N ILE A 61 20.717 22.964 12.260 1.00 11.08 N +ATOM 476 CA ILE A 61 21.184 24.263 11.690 1.00 11.78 C +ATOM 477 C ILE A 61 21.110 24.111 10.173 1.00 13.74 C +ATOM 478 O ILE A 61 21.841 23.198 9.686 1.00 14.60 O +ATOM 479 CB ILE A 61 22.650 24.516 12.172 1.00 11.80 C +ATOM 480 CG1 ILE A 61 22.662 24.819 13.699 1.00 11.56 C +ATOM 481 CG2 ILE A 61 23.376 25.645 11.409 1.00 13.29 C +ATOM 482 CD1 ILE A 61 24.123 24.981 14.195 1.00 11.42 C +ATOM 483 N GLN A 62 20.291 24.875 9.507 1.00 13.97 N +ATOM 484 CA GLN A 62 20.081 24.773 8.033 1.00 15.52 C +ATOM 485 C GLN A 62 20.822 25.914 7.332 1.00 13.94 C +ATOM 486 O GLN A 62 21.323 26.830 8.008 1.00 12.15 O +ATOM 487 CB GLN A 62 18.599 24.736 7.727 1.00 19.53 C +ATOM 488 CG GLN A 62 17.819 23.434 7.900 1.00 26.38 C +ATOM 489 CD GLN A 62 16.509 23.529 7.116 1.00 30.61 C +ATOM 490 OE1 GLN A 62 15.446 22.980 7.433 1.00 33.23 O +ATOM 491 NE2 GLN A 62 16.539 24.293 6.009 1.00 32.71 N +ATOM 492 N LYS A 63 20.924 25.862 6.006 1.00 11.73 N +ATOM 493 CA LYS A 63 21.656 26.847 5.240 1.00 11.97 C +ATOM 494 C LYS A 63 21.127 28.240 5.574 1.00 10.41 C +ATOM 495 O LYS A 63 19.958 28.465 5.842 1.00 9.59 O +ATOM 496 CB LYS A 63 21.631 26.642 3.731 1.00 13.73 C +ATOM 497 CG LYS A 63 20.210 26.423 3.175 1.00 16.98 C +ATOM 498 CD LYS A 63 20.268 26.589 1.656 1.00 20.19 C +ATOM 499 CE LYS A 63 19.202 25.857 0.891 1.00 23.42 C +ATOM 500 NZ LYS A 63 17.884 26.544 1.075 1.00 25.97 N +ATOM 501 N GLU A 64 22.099 29.163 5.605 1.00 10.04 N +ATOM 502 CA GLU A 64 21.907 30.563 5.881 1.00 10.94 C +ATOM 503 C GLU A 64 21.466 30.953 7.261 1.00 9.74 C +ATOM 504 O GLU A 64 21.066 32.112 7.533 1.00 9.42 O +ATOM 505 CB GLU A 64 21.023 31.223 4.784 1.00 18.31 C +ATOM 506 CG GLU A 64 21.861 31.342 3.474 1.00 24.16 C +ATOM 507 CD GLU A 64 21.156 30.726 2.311 1.00 29.00 C +ATOM 508 OE1 GLU A 64 19.942 30.793 2.170 1.00 31.72 O +ATOM 509 OE2 GLU A 64 21.954 30.152 1.535 1.00 32.61 O +ATOM 510 N SER A 65 21.674 30.034 8.191 1.00 6.85 N +ATOM 511 CA SER A 65 21.419 30.253 9.620 1.00 6.90 C +ATOM 512 C SER A 65 22.504 31.228 10.136 1.00 4.72 C +ATOM 513 O SER A 65 23.579 31.321 9.554 1.00 3.91 O +ATOM 514 CB SER A 65 21.637 28.923 10.353 1.00 7.28 C +ATOM 515 OG SER A 65 20.544 28.047 10.059 1.00 10.56 O +ATOM 516 N THR A 66 22.241 31.873 11.241 1.00 4.48 N +ATOM 517 CA THR A 66 23.212 32.762 11.891 1.00 3.80 C +ATOM 518 C THR A 66 23.509 32.224 13.290 1.00 4.60 C +ATOM 519 O THR A 66 22.544 31.942 14.034 1.00 5.33 O +ATOM 520 CB THR A 66 22.699 34.267 11.985 1.00 2.85 C +ATOM 521 OG1 THR A 66 22.495 34.690 10.589 1.00 2.15 O +ATOM 522 CG2 THR A 66 23.727 35.131 12.722 1.00 3.40 C +ATOM 523 N LEU A 67 24.790 32.021 13.618 1.00 4.17 N +ATOM 524 CA LEU A 67 25.149 31.609 14.980 1.00 3.85 C +ATOM 525 C LEU A 67 25.698 32.876 15.669 1.00 3.80 C +ATOM 526 O LEU A 67 26.158 33.730 14.894 1.00 5.54 O +ATOM 527 CB LEU A 67 26.310 30.594 14.967 1.00 7.18 C +ATOM 528 CG LEU A 67 26.290 29.480 13.960 1.00 9.67 C +ATOM 529 CD1 LEU A 67 27.393 28.442 14.229 1.00 8.12 C +ATOM 530 CD2 LEU A 67 24.942 28.807 13.952 1.00 11.66 C +ATOM 531 N HIS A 68 25.621 32.945 16.950 1.00 2.94 N +ATOM 532 CA HIS A 68 26.179 34.127 17.650 1.00 4.17 C +ATOM 533 C HIS A 68 27.475 33.651 18.304 1.00 5.32 C +ATOM 534 O HIS A 68 27.507 32.587 18.958 1.00 7.70 O +ATOM 535 CB HIS A 68 25.214 34.565 18.780 1.00 5.57 C +ATOM 536 CG HIS A 68 23.978 35.121 18.126 1.00 9.95 C +ATOM 537 ND1 HIS A 68 23.853 36.432 17.781 1.00 13.74 N +ATOM 538 CD2 HIS A 68 22.824 34.514 17.782 1.00 12.79 C +ATOM 539 CE1 HIS A 68 22.674 36.627 17.200 1.00 14.75 C +ATOM 540 NE2 HIS A 68 22.045 35.455 17.173 1.00 16.30 N +ATOM 541 N LEU A 69 28.525 34.447 18.189 1.00 5.29 N +ATOM 542 CA LEU A 69 29.801 34.145 18.829 1.00 3.97 C +ATOM 543 C LEU A 69 30.052 35.042 20.004 1.00 5.07 C +ATOM 544 O LEU A 69 30.105 36.305 19.788 1.00 4.34 O +ATOM 545 CB LEU A 69 30.925 34.304 17.753 1.00 6.08 C +ATOM 546 CG LEU A 69 32.345 34.183 18.358 1.00 7.37 C +ATOM 547 CD1 LEU A 69 32.555 32.783 18.870 1.00 6.87 C +ATOM 548 CD2 LEU A 69 33.361 34.491 17.245 1.00 9.96 C +ATOM 549 N VAL A 70 30.124 34.533 21.191 1.00 4.29 N +ATOM 550 CA VAL A 70 30.479 35.369 22.374 1.00 6.26 C +ATOM 551 C VAL A 70 31.901 34.910 22.728 1.00 9.22 C +ATOM 552 O VAL A 70 32.190 33.696 22.635 1.00 9.36 O +ATOM 553 CB VAL A 70 29.472 35.181 23.498 1.00 8.69 C +ATOM 554 CG1 VAL A 70 29.821 35.957 24.765 1.00 9.76 C +ATOM 555 CG2 VAL A 70 28.049 35.454 23.071 1.00 8.54 C +ATOM 556 N LEU A 71 32.763 35.831 23.090 1.00 12.71 N +ATOM 557 CA LEU A 71 34.145 35.472 23.481 1.00 16.06 C +ATOM 558 C LEU A 71 34.239 35.353 24.979 1.00 18.09 C +ATOM 559 O LEU A 71 33.707 36.197 25.728 1.00 19.26 O +ATOM 560 CB LEU A 71 35.114 36.564 22.907 1.00 17.10 C +ATOM 561 CG LEU A 71 35.926 35.979 21.737 1.00 19.37 C +ATOM 562 CD1 LEU A 71 35.003 35.084 20.920 1.00 17.51 C +ATOM 563 CD2 LEU A 71 36.533 37.087 20.917 1.00 19.57 C +ATOM 564 N ARG A 72 34.930 34.384 25.451 1.00 21.47 N +ATOM 565 CA ARG A 72 35.161 34.174 26.896 1.00 25.83 C +ATOM 566 C ARG A 72 36.671 34.296 27.089 1.00 27.74 C +ATOM 567 O ARG A 72 37.305 33.233 26.795 1.00 30.65 O +ATOM 568 CB ARG A 72 34.717 32.760 27.286 1.00 28.49 C +ATOM 569 CG ARG A 72 35.752 32.054 28.160 1.00 31.79 C +ATOM 570 CD ARG A 72 35.612 30.577 28.044 1.00 34.05 C +ATOM 571 NE ARG A 72 35.040 30.252 26.730 1.00 35.08 N +ATOM 572 CZ ARG A 72 34.338 29.103 26.650 1.00 34.67 C +ATOM 573 NH1 ARG A 72 34.110 28.437 27.768 1.00 35.02 N +ATOM 574 NH2 ARG A 72 34.014 28.657 25.457 1.00 34.97 N +ATOM 575 N LEU A 73 37.197 35.397 27.513 0.45 28.93 N +ATOM 576 CA LEU A 73 38.668 35.502 27.680 0.45 30.76 C +ATOM 577 C LEU A 73 39.076 34.931 29.031 0.45 32.18 C +ATOM 578 O LEU A 73 38.297 34.946 29.996 0.45 32.31 O +ATOM 579 CB LEU A 73 39.080 36.941 27.406 0.45 30.53 C +ATOM 580 CG LEU A 73 39.502 37.340 26.002 0.45 30.16 C +ATOM 581 CD1 LEU A 73 38.684 36.647 24.923 0.45 29.57 C +ATOM 582 CD2 LEU A 73 39.337 38.854 25.862 0.45 29.11 C +ATOM 583 N ARG A 74 40.294 34.412 29.045 0.45 33.82 N +ATOM 584 CA ARG A 74 40.873 33.802 30.253 0.45 35.33 C +ATOM 585 C ARG A 74 41.765 34.829 30.944 0.45 36.22 C +ATOM 586 O ARG A 74 42.945 34.994 30.583 0.45 36.70 O +ATOM 587 CB ARG A 74 41.651 32.529 29.923 0.45 36.91 C +ATOM 588 CG ARG A 74 41.608 31.444 30.989 0.45 38.62 C +ATOM 589 CD ARG A 74 41.896 30.080 30.456 0.45 39.75 C +ATOM 590 NE ARG A 74 43.311 29.735 30.563 0.45 41.13 N +ATOM 591 CZ ARG A 74 44.174 29.905 29.554 0.45 41.91 C +ATOM 592 NH1 ARG A 74 43.754 30.312 28.356 0.45 42.75 N +ATOM 593 NH2 ARG A 74 45.477 29.726 29.763 0.45 41.93 N +ATOM 594 N GLY A 75 41.165 35.531 31.898 0.25 36.31 N +ATOM 595 CA GLY A 75 41.845 36.550 32.686 0.25 36.07 C +ATOM 596 C GLY A 75 41.251 37.941 32.588 0.25 36.16 C +ATOM 597 O GLY A 75 41.102 38.523 31.500 0.25 36.26 O +ATOM 598 N GLY A 76 40.946 38.472 33.757 0.25 36.05 N +ATOM 599 CA GLY A 76 40.373 39.813 33.944 0.25 36.19 C +ATOM 600 C GLY A 76 40.031 39.992 35.432 0.25 36.20 C +ATOM 601 O GLY A 76 38.933 40.525 35.687 0.25 36.13 O +ATOM 602 OXT GLY A 76 40.862 39.575 36.251 0.25 36.27 O +TER 603 GLY A 76 +END diff --git a/tools/tmol/test/check.py b/tools/tmol/test/check.py new file mode 100755 index 0000000..ce365d0 --- /dev/null +++ b/tools/tmol/test/check.py @@ -0,0 +1,193 @@ +#!/usr/bin/env python3 +"""Check tmol's test outputs against the known fixture. + +The fixture is water-stripped 1UBQ (76 residues; the record's 58 waters +removed) in 26 copies. 26 crosses tmol's first input-growth threshold: +the packer chunks a pose stack whose poses hold at most 256 blocks into +chunks of 25 poses, so `tmol pack` on this batch runs the chunked path -- +25 poses, then 1. Fewer than 25 poses would never leave one chunk. + +The reference values are pinned by the campaign's measurement run on +this same fixture (NVIDIA L40S, driver 580.178.04; torch 2.14.1+cu126, +tmol 1.0.0 with tmol-ops-cu12 1.0.0 -- the image's pins), which also +measured what each tolerance has to cover: + +- score: the 26 identical poses spread 0.000274 within one batch, and + two runs in separate processes differ by at most 0.000183 -- GPU + score reductions are deterministic only to that noise, never byte + for byte, so the checks below compare within it and + TOTAL_TOLERANCE keeps 100x headroom for a different GPU; +- minimize lowered every pose by at least 437.3, pack by 288.2, relax + by 473.0, and the demo script by 473.9; +- two `tmol pack --seed 1` runs wrote byte-identical structures, their + scores.tsv totals 0.000161 apart; +- `tmol score` on minimize's outdir reproduced its scores.tsv within + 0.000184 -- the same reduction noise, so the chain check tolerates + it instead of demanding byte equality. +""" +import csv +import math +import sys +from pathlib import Path + +T = Path(sys.argv[1]) + +NPOSES = 26 +NAMES = [f"pose_{i:02d}.pdb" for i in range(1, NPOSES + 1)] + +# Pinned by the measurement run described above. +HEADER = ( # a score TSV's exact header: name, total, 25 weighted terms + "name\ttotal\tfa_ljatr\tfa_ljrep\tfa_lk\tfa_elec\thbond\t" + "cart_lengths\tcart_angles\tcart_torsions\tcart_impropers\t" + "cart_hxltorsions\tdisulfide\trama\tomega\tdunbrack_rot\t" + "dunbrack_rotdev\tdunbrack_semirot\tlk_ball_iso\tlk_ball\t" + "lk_bridge\tlk_bridge_uncpl\tref\tgen_torsions\tna_torsion\t" + "na_torsion_well\tmetal_coordination" +) +REFERENCE_TOTAL = 195.064756 # the batch's mean total; poses spread around it +TOTAL_TOLERANCE = 0.05 # cross-run noise is 0.0004; headroom for another GPU +POSE_SPREAD = 0.01 # identical poses in one batch spread 0.000274 +MIN_LOWER_BY = 400 # measured: every pose lowered by at least 437.3 +PACK_LOWER_BY = 250 # measured: at least 288.2 +RELAX_LOWER_BY = 400 # measured: at least 473.0 +DEMO_LOWER_BY = 400 # measured: 473.9 +SEEDED_MAX_TOTAL_DELTA = 0.01 # the seeded pair measured 0.000161 +SEEDED_MAX_COORD_DELTA = 0.05 # the pair measured 0.0; a flipped rotamer is > 0.5 +CHAIN_MAX_TOTAL_DELTA = 0.01 # the chain measured 0.000184, not byte-identical + + +def fail(msg): + raise SystemExit(f"FAIL: {msg}") + + +def rows(tsv): + with open(tsv) as handle: + return list(csv.DictReader(handle, delimiter="\t")) + + +def totals_of(tsv): + return [float(row["total"]) for row in rows(tsv)] + + +def check_scores_tsv(path, label): + header = path.read_text().splitlines()[0] + if header != HEADER: + fail(f"{label}: header is not the pinned beta2016 column set") + table = rows(path) + if len(table) != NPOSES: + fail(f"{label}: {len(table)} rows, expected {NPOSES}") + if [row["name"] for row in table] != NAMES: + fail(f"{label}: rows are not the inputs, in order, named for them") + totals = [float(row["total"]) for row in table] + if not all(math.isfinite(total) for total in totals): + fail(f"{label}: a total is not finite") + spread = max(totals) - min(totals) + if spread > POSE_SPREAD: + fail(f"{label}: identical poses spread {spread:.6f} > {POSE_SPREAD}") + worst = max(abs(total - REFERENCE_TOTAL) for total in totals) + if worst > TOTAL_TOLERANCE: + fail(f"{label}: total differs from the pinned reference by " + f"{worst:.6f} > {TOTAL_TOLERANCE}") + return totals + + +def check_verb_outdir(directory, label, lowered_by): + scores = directory / "scores.tsv" + if not scores.is_file(): + fail(f"{label}: {scores} is missing") + structures = sorted(p.name for p in directory.iterdir() + if p.name != "scores.tsv") + if structures != NAMES: + fail(f"{label}: not one written structure per input, named for it") + totals = totals_of(scores) + if not all(math.isfinite(total) for total in totals): + fail(f"{label}: a total is not finite") + least = min(REFERENCE_TOTAL - total for total in totals) + if least < lowered_by: + fail(f"{label}: a pose was lowered by only {least:.3f} < {lowered_by}") + + +def pdb_coords(path): + values = [] + for line in path.read_text().splitlines(): + if line.startswith(("ATOM ", "HETATM")): + values.append((float(line[30:38]), float(line[38:46]), + float(line[46:54]))) + return values + + +def check_seeded_pair(): + a = {p.name: p.read_bytes() for p in (T / "pack-a").iterdir()} + b = {p.name: p.read_bytes() for p in (T / "pack-b").iterdir()} + if a == b: + return # byte-identical: the strongest outcome + if set(a) != set(b): + fail("seeded pack pair: different file sets") + totals_a = totals_of(T / "pack-a" / "scores.tsv") + totals_b = totals_of(T / "pack-b" / "scores.tsv") + worst_total = max(abs(x - y) for x, y in zip(totals_a, totals_b)) + worst_coord = 0.0 + for name in NAMES: + coords_a = pdb_coords(T / "pack-a" / name) + coords_b = pdb_coords(T / "pack-b" / name) + if len(coords_a) != len(coords_b): + fail(f"seeded pack pair: {name} atom counts differ") + worst_coord = max( + [worst_coord] + [abs(x - y) + for pa, pb in zip(coords_a, coords_b) + for x, y in zip(pa, pb)] + ) + if worst_total > SEEDED_MAX_TOTAL_DELTA: + fail(f"seeded pack pair: totals differ by {worst_total:.6f} " + f"> {SEEDED_MAX_TOTAL_DELTA}") + if worst_coord > SEEDED_MAX_COORD_DELTA: + fail(f"seeded pack pair: coordinates differ by " + f"{worst_coord:.6f} > {SEEDED_MAX_COORD_DELTA} A") + + +def check_chain(): + written = (T / "min" / "scores.tsv").read_bytes() + rescored = (T / "score-min.tsv").read_bytes() + if written == rescored: + return # byte-identical: the strongest outcome + # Not byte-identical on GPU: score reductions carry ~1e-4 of noise + # between processes. The written and rescored tables must still hold + # the same totals, pose by pose. + worst = max(abs(x - y) for x, y in zip( + totals_of(T / "min" / "scores.tsv"), totals_of(T / "score-min.tsv"))) + if worst > CHAIN_MAX_TOTAL_DELTA: + fail("tmol score on minimize's outdir does not reproduce scores.tsv " + f"within {CHAIN_MAX_TOTAL_DELTA}: {worst:.6f}") + + +def check_demo(): + lines = (T / "demo.txt").read_text().splitlines() + if (len(lines) != 2 or not lines[0].startswith("before ") + or not lines[1].startswith("after ")): + fail("demo: expected 'before ' and 'after ' lines") + before = float(lines[0].split()[1]) + after = float(lines[1].split()[1]) + if abs(before - REFERENCE_TOTAL) > TOTAL_TOLERANCE: + fail("demo: the single-pose score is off the batch reference") + if REFERENCE_TOTAL - after < DEMO_LOWER_BY: + fail(f"demo: relax lowered the score by only " + f"{REFERENCE_TOTAL - after:.3f} < {DEMO_LOWER_BY}") + + +def main(): + totals = check_scores_tsv(T / "scores.tsv", "score") + check_verb_outdir(T / "min", "minimize", MIN_LOWER_BY) + check_verb_outdir(T / "pack-a", "pack", PACK_LOWER_BY) + check_verb_outdir(T / "relax", "relax", RELAX_LOWER_BY) + check_seeded_pair() + check_chain() + check_demo() + if not (T / "demo-relaxed.pdb").is_file(): + fail("demo: the relaxed structure was not written") + print(f"tmol test OK: 26 poses of 1UBQ at {totals[0]:.3f}; " + "minimize/pack/relax each lowered it; the seeded pair " + "reproduces; scores.tsv chains through tmol score") + + +if __name__ == "__main__": + main() diff --git a/tools/tmol/test/run.sh b/tools/tmol/test/run.sh new file mode 100755 index 0000000..ca082ae --- /dev/null +++ b/tools/tmol/test/run.sh @@ -0,0 +1,68 @@ +#!/usr/bin/env bash +# tmol's test. The fixture is water-stripped 1UBQ (76 residues; the +# record's 58 waters removed), expanded to 26 copies: tmol's packer +# processes a pose stack whose poses hold at most 256 blocks in chunks +# of 25 poses, so 26 copies crosses the boundary and the test exercises +# the chunked path (25 poses, then 1). Every documented command runs: the +# four verbs on the directory batch, a seeded reproducibility pair, the +# chaining invariant (tmol score on an optimizer verb's outdir +# reproduces its scores.tsv), and the library way (one python3 -I +# script that relaxes then scores, in one process). check.py verifies +# the answers against values pinned by the campaign's measurement run. +set -euo pipefail +cd "$(dirname "$0")" +T=$(mktemp -d /tmp/tmol.XXXXXX) +trap 'rm -rf "$T"' EXIT + +mkdir -p "$T/in" +for i in $(seq -w 1 26); do + cp 1UBQ.pdb "$T/in/pose_$i.pdb" +done + +tmol score "$T/in" "$T/scores.tsv" +tmol minimize "$T/in" "$T/min" +tmol pack "$T/in" "$T/pack-a" --seed 1 +tmol pack "$T/in" "$T/pack-b" --seed 1 +tmol relax "$T/in" "$T/relax" + +# the chaining invariant: scoring a verb's outdir reproduces its scores.tsv +tmol score "$T/min" "$T/score-min.tsv" + +# the library way: relax then score, one process, under python3 -I +cat > "$T/relax_score.py" <<'PY' +import sys + +import torch +from tmol import ( + beta2016_score_function, + fast_relax, + file_from_pose_stack, + pose_stack_from_file, +) + +source, output = sys.argv[1], sys.argv[2] +device = torch.device("cuda" if torch.cuda.is_available() else "cpu") +torch.manual_seed(0) + +pose = pose_stack_from_file(source, device, packer_seed=0) +sfxn = beta2016_score_function(device) +scorer = sfxn.render_whole_pose_scoring_module(pose) +with torch.no_grad(): + before = scorer(pose.coords).tolist() + +relaxed = fast_relax(pose, sfxn) + +# Re-render the scorer on the relaxed pose: its atom layout can change, +# and the old scorer is bound to the old layout. +scorer = sfxn.render_whole_pose_scoring_module(relaxed) +with torch.no_grad(): + after = scorer(relaxed.coords).tolist() + +file_from_pose_stack(relaxed, output) +print(f"before {before[0]:.6f}") +print(f"after {after[0]:.6f}") +PY +python3 -I "$T/relax_score.py" "$T/in/pose_01.pdb" "$T/demo-relaxed.pdb" \ + > "$T/demo.txt" + +python3 check.py "$T"