spiralgenetics / biograph Goto Github PK

vdb study list and vdb study show should indicate if a study is frozen

Hi,

Could you provide a download link for the biograph model (https://archive.spiralgenetics.com/files/models/biograph_model-7.1.0.ml) the current one is broken? I am happy to upload this file on a reliable storage platform.

I could not locate this file in the docker container.

Thanks,
Yann

PyVCF can only be installed with use_2to3, which is no longer used in any setuptools past 58. Therefore, need to remove PyVCF dependency.

PyVCF is used:

• ./python/biograph/tools/coverage.py
  • Just uses it for header stuff
• ./python/functest/biograph_test.py
  • Single line just to count vcf entries
• ./python/functest/GenomeGraphTests.py
  • Again, just a counter

There are a few places in ./python/biograph/internal/ where PyVCF is also used. But before refactoring that code, we should check if we even need to support those tools at all.

I can't reproduce this error outside of github-ci, but it fails fairly regularly when testing with or without -copt.

https://github.com/spiralgenetics/biograph/runs/3624148617?check_suite_focus=true

Trying to export a study without a properly set TMPDIR gave error:

Traceback (most recent call last):File "/usr/lib/python3.6/multiprocessing/process.py", line 258, in _bootstrapself.run()
File "/home/ubuntu/pyenv/lib/python3.6/site-packages/biograph/vdb/study_cmd.py", line 97, in runfile=fOSError: [Errno 28] No space left on deviceExporting VCF

So it failed, and then kept going, only exporting chromosome 1 and part of chromosome 10

Every time a new connection is made to athena or s3, boto reports:

Found credentials in shared credentials file: ~/.aws/credentials

This is quite annoying when running many child processes in parallel.

Implement a logging filter, eg. https://stackoverflow.com/questions/879732/logging-with-filters

Implement phenotype tagging for sample names, and cohort selection based on phenotype criteria.

It appears that v7.1.0 only included _capi_36.so. We should cut a new minor release with support for 3.6-3.8.

Create arbitrary groups of aids and use the cohort identifiers in place of any command that requires sample names or aids.

Since there is only one sample exported, N_MISS is always total_missing - 1.

Either calculate it correctly for the whole cohort or don’t report it when running --square-off.

Would be nice to be able to export subset by regions. bcftools allows a -r chrom:start-end as well as a -r regions.bed.

Is biograph able to joint-genotype across a set of project samples? How does one create a project level VCF for analyses with biographies?