Add kegalign and ucsc_chainmergesort to the Mapping section - #1727
Open
richard-burhans wants to merge 7 commits into
Open
richard-burhans wants to merge 7 commits into
richard-burhans wants to merge 7 commits into
Conversation
This revision adds two tools to the repository alongside kegalign 0.3.2+galaxy0: batched_lastz 1.04.52+galaxy3 and growler_lastz 1.04.52+galaxy1. The three share a scoring contract and a file format and were previously maintained in separate Tool Shed repositories.
added 2 commits
September 19, 2026 09:26
Revision 9fdb7d427703 (numeric 19) is no longer the latest: 49249635e892 (numeric 20) was published on 2026-09-19 and carries the change this install is most wanted for. batched_lastz concatenates every per-command alignment into one gzip stream at the end of its job, and that single write is ~62% of the job's wall clock. Revision 19 does it at the zlib default of 6; revision 20 at level 1. Measured on 1.69 GB of real alignment output: 18 MB/s against 137 MB/s. Both revisions are listed, as the section already lists eleven others.
The two revisions were both listed. Revision 9fdb7d427703 (numeric 19) has never been installed anywhere -- this pull request is what would have installed it -- so there is no history or workflow depending on it, and 49249635e892 (numeric 20) supersedes it in every tool: rev 19 batched_lastz 1.04.52+galaxy3 growler_lastz +galaxy1 kegalign 0.3.2+galaxy0 rev 20 batched_lastz 1.04.52+galaxy4 growler_lastz +galaxy2 kegalign 0.3.2+galaxy1 The reason to carry old revisions is that removing an installed one breaks the histories and workflows that ran against it. Neither applies here: usegalaxy.org currently has kegalign 0.3.1+galaxy0 and batched_lastz 1.04.22+galaxy8, and growler_lastz not at all. Listing 19 would install three tool versions nobody has ever used, one of which does its final gzip at level 6 rather than level 1.
chainMergeSort had no Galaxy tool anywhere -- absent from tools-iuc, which wraps nine other UCSC chain tools, from the Tool Shed, and from this file. The Growler per-pair wrapper cannot run without it: it is where the chromosome-pair axis collapses back into one score-ordered stream, and chainPreNet and chainNet cannot be partitioned by chromosome because each walks one stream marking target and query space simultaneously. It is a SEPARATE shed repository from kegalign, so the kegalign entry in this file does not cover it. ucsc_chainmergesort richard-burhans rev 9e355f0c9f46 482+galaxy0 The collapse it performs is measured exact against chaining a genome pair whole: 28 real Cannabis chromosome pairs, 17,696 chains, identical chain bodies. And its renumbering is load-bearing -- -saveId leaves 15,691 colliding chain ids out of those 17,696, which netChainSubset resolves by id.
richard-burhans
changed the title
added 2 commits
September 19, 2026 10:09
Swaps 49249635e892 (numeric 20) for e288740b966a (numeric 22). Neither
has ever been installed, so nothing depends on the older one, and 22
supersedes it in every tool:
rev 20 batched_lastz 1.04.52+galaxy4 growler_lastz +galaxy2 kegalign 0.3.2+galaxy1
rev 22 batched_lastz 1.04.52+galaxy6 growler_lastz +galaxy3 kegalign 0.3.2+galaxy2
The difference that matters is batched_lastz. Its final step concatenates
every per-command alignment into one gzip stream, and that single write
is ~62% of the job's wall clock -- measured twice, on inputs differing
9x. Revision 20 does it with Python's gzip at level 1; revision 22 pipes
it through pigz across the job's whole slot budget, which is idle by
then because the compute workers have finished.
Measured on 1.69 GB of real alignment output:
the original, gzip -6 18 MB/s
revision 20, gzip -1 137 MB/s
revision 22, pigz -1 2456 MB/s standalone, 997 MB/s end to end
through the tool, 6.6x over revision 20
Installing 20 as well would add three tool versions nobody has used, one
of which is strictly slower at the step that dominates the job.
scripts/fix_lockfile.py takes the SOURCE .yml and writes the .lock beside it -- the Makefile feeds it `find -name '*.yml'`. I ran it on the .lock file instead, so it treated that as a source and generated mapping.yml.lock.lock, 322 lines of duplicate. ⚠ Which also means the three times I ran it "to validate" the lock validated nothing: it was generating a new file each time, not checking the one I had edited. The check it should have been is `fix_lockfile.py usegalaxy.org/mapping.yml` followed by a diff, and that now passes: the generated lock is byte-identical to the hand edit, so both entries -- ucsc_chainmergesort and the kegalign revision swap -- are exactly what the tooling produces.
richard-burhans
enabled auto-merge (squash)
September 19, 2026 17:32
Revision 22 was pinned before KegAlign v0.3.3 existed. Revision 23
published while this PR was open and is a strict improvement, so this
swaps rather than adds -- there is no reason to install 22 at all.
Revision 23 carries all three tools in the repository:
kegalign 0.3.3+galaxy0
batched_lastz 1.04.52+galaxy7
growler_lastz 1.04.52+galaxy4
0.3.2's container ships the buffered build_pairs.py, whose measured peak
RSS is 52.4 MB + 1.96 x (input MB) -- about 28 GB for a 14.15 GB bundle,
and OOM-killed on a 15 GB machine. The tool invokes exactly that path.
0.3.3 is the first release whose collection output survives a real
genome pair.
All three containers resolve today:
kegalign-full:0.3.3--hdfd78af_0
mulled-v2-22e8b6fd71e23314d99ed252d036fc77e59901b5:528c639a...-0
lastz:1.04.52--h7b50bb2_1
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Member
Author
|
Repointed from kegalign revision 22 ( Revision 23 carries all three tools in the repository:
Why 0.3.3 rather than 0.3.2. 0.3.2's container ships the buffered Measured peak RSS for that path is 52.4 MB + 1.96 × (input MB) — ~28 GB extrapolated for a 14.15 GB bundle, and OOM-killed on a 15 GB machine. The tool invokes exactly that path ( All three containers resolve today: (The Diff is one line swapped for one line; 🤖 Generated with Claude Code |
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Installs Tool Shed revision
9fdb7d427703ofrichard-burhans/kegalignon usegalaxy.org.This revision carries three tools rather than one.
batched_lastzandgrowler_lastzpreviouslylived in their own Tool Shed repositories; they share a scoring contract, a macro file and a file
format with
kegalign, so they have been merged into a single repository. Installing this revisiontherefore adds two tools alongside the
kegalignupgrade:kegalignbatched_lastzgrowler_lastzkegalign0.3.2 adds an optional collection output: instead of a tarball of LASTZ commands, it canemit one gzipped segment file per chromosome pair plus the two
.2bitfiles.growler_lastzis theconsumer for that shape, mapped over the collection. The existing tarball path is unchanged and
remains the default.
Superseded revisions that were never propagated here are skipped, per this repository's convention
of installing the latest installable revision only.
Installation sequence for
tool-installers