Fails to load BED file

[olegs]

I am trying to load the bed file (see attachement GSM646316_GM12878_H3K27ac_rep1_peaks.bed.zip
)

import pyranges as pr
file = 'GSM646316_GM12878_H3K27ac_rep1_peaks.bed'
pr.read_bed(file)

# This works:
import pandas as pd
pd.read_csv(file, sep='\t', header=False)

it fails with the error:
Length mismatch: Expected axis has 14 elements, new values have 12 elements, however pandas successfully loads the file.

Version of pyranges installed with conda, bioconda channel. Python 3.10.12, Ubuntu 22.04 x86_64
pyranges 0.1.4 pyhdfd78af_0 bioconda

[marco-mariotti]

Hi, do you mind trying with pyranges v1? If it still fails, please open an issue there too, just with the link to this issue https://github.com/pyranges/pyranges_1.x

…

On Tue, Sep 30, 2025 at 10:11 AM Oleg Shpynov ***@***.***> wrote: *olegs* created an issue (pyranges/pyranges0#400) <#400> I am trying to load the bed file (see attachement GSM646316_GM12878_H3K27ac_rep1_peaks.bed.zip <https://github.com/user-attachments/files/22613766/GSM646316_GM12878_H3K27ac_rep1_peaks.bed.zip> ) import pyranges as pr file = 'GSM646316_GM12878_H3K27ac_rep1_peaks.bed' pr.read_bed(file) # This works: import pandas as pd pd.read_csv(file, sep='\t', header=False) it fails with the error: Length mismatch: Expected axis has 14 elements, new values have 12 elements, however pandas successfully loads the file. Version of pyranges installed with conda, bioconda channel. Python 3.10.12, Ubuntu 22.04 x86_64 pyranges 0.1.4 pyhdfd78af_0 bioconda — Reply to this email directly, view it on GitHub <#400>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/ACTXRIHRZN3KZKHMJTSEF5L3VI3KHAVCNFSM6AAAAACH33HT46VHI2DSMVQWIX3LMV43ASLTON2WKOZTGQ3DQMBZG4ZDOMA> . You are receiving this because you are subscribed to this thread.Message ID: ***@***.***>

-- Marco Mariotti Ramón y Cajal Fellow, Group Leader Department of Genetics, Microbiology and Statistics Universitat de Barcelona

[endrebak]

This might be because bed has a specific meaning. It is a tabular file with 3, 4, 6, 12 columns IIRC. Lemme check

[endrebak]

ChatGPT:

1. What a BED file is supposed to look like

• Required:

  • column 1 = chrom
  • column 2 = chromStart (0-based, inclusive)
  • column 3 = chromEnd (0-based, exclusive)

• Optional (in order):
  4 = name, 5 = score (0–1000), 6 = strand (+/-),
  7 = thickStart, 8 = thickEnd, 9 = itemRgb,
  10 = blockCount, 11 = blockSizes, 12 = blockStarts.

A plain “BED3” needs 3 columns; “BED6” has 6; “BED9” has 9; “BED12” has 12.
If you supply more than 12 columns there is no standard specification — tools will typically either ignore extra fields or error out.

---

2. Your line

chr1    778412  778642  +  15.6  0.800  11.000000  8.9  1.0  8.89  1.02e-05  9.12  6.54e-08  0.89

Count: 14 fields.

• Col 1–3: chr1, 778412, 778642 → OK for BED3.
• Col 4: "+" → would normally be name, but here it’s a strand symbol, so already not matching the official order.
• Col 5: 15.6 → would normally be score (must be 0–1000 integer if used by UCSC). Here it’s a float.
• Col 6: 0.800 → no defined meaning in BED spec.
• And you have 14 total, whereas the spec tops out at 12.

---

3. Conclusion

• Strictly speaking: No — this is not a valid BED file under the UCSC BED specification.

  • Wrong column order (strand is in col 4, should be col 6).
  • Non-integer “score”.
  • More than the maximum 12 defined fields.

• Practically: many tools (e.g. bedtools, pybedtools, pyranges) will happily read “BED-like” tab-separated coordinates if you only need chrom/start/end, because they ignore or just pass through extra columns. But anything expecting canonical BED6/BED12 will reject or misinterpret this.

---

Tip:
If you want it spec-conformant, either cut it to BED3/6:

cut -f1-6 GSM646316_GM12878_H3K27ac_rep1_peaks.bed > peaks.bed6

(or rename/reorder columns to match name,score,strand,... if you need BED6+).

``

[endrebak]

I guess a better error message is in order