Feature suggestion: accept comma separated lists of overlapping genes for plots

[etal]

Feature suggestion: accept comma separated lists of overlapping genes for plots

[etal]

Does this mean a single row/bin/probe in the .cnr or .cnn file has a gene name like "FOO,BAR"?

Normally the scatter command will take "-g FOO,BAR" and look for a gene named FOO and another gene named BAR on the same chromosome. So it will not find the bins labeled "FOO,BAR" -- is that the bug? If so, I'm not sure what the proper fix is -- maybe if the search for "FOO" and "BAR" fails, try again with "FOO,BAR". Changing the delimiter used for "-g" isn't right, because any separator could be used to join the names of overlapping genes in the BED file.

Workaround: Replace the gene name separator in your BED/.cnn/.cnr/.cns files with something other than , -- like ; instead:

sed -i "s/,/;/g" my_sample.cnr

(Or have bedtools use a different separator, if that's how the original BED file of genes was created.)

[mjafin]

That's it, thanks, will give it a try!

[chapmanb]

Eric and Miika;
Thanks a lot for the discussion and workaround. I swapped this over in bcbio to use ';' which will hopefully give us compatibility with CNVkit automatically. Miika, let us know if this doesn't fix it for you.

[mjafin]

Thanks Eric and Brad, I'll test all of this and will reopen if I come across any issues.

[mjafin]

@etal So we replaced commas by semicolons like this:
chr14 23440198 23441327 AJUBA;RP11-298I3.5 -0.19098 1.0
However I'm still getting this error:

AURA_142
Traceback (most recent call last):
  File "/group/ngs/src/bcbio-nextgen/0.8.8/rhel6-x64/anaconda/bin/cnvkit.py", line 4, in <module>
    __import__('pkg_resources').run_script('CNVkit==0.3.5.dev0', 'cnvkit.py')
  File "/group/ngs/src/bcbio-nextgen/0.8.8/rhel6-x64/anaconda/lib/python2.7/site-packages/setuptools-15.0-py2.7.egg/pkg_resources/__init__.py", line 723, in run_script
  File "/group/ngs/src/bcbio-nextgen/0.8.8/rhel6-x64/anaconda/lib/python2.7/site-packages/setuptools-15.0-py2.7.egg/pkg_resources/__init__.py", line 1643, in run_script
  File "/group/ngs/src/bcbio-nextgen/0.8.8/rhel6-x64/anaconda/lib/python2.7/site-packages/CNVkit-0.3.5.dev0-py2.7.egg/EGG-INFO/scripts/cnvkit.py", line 9, in <module>

  File "build/bdist.linux-x86_64/egg/cnvlib/commands.py", line 719, in _cmd_scatter

  File "build/bdist.linux-x86_64/egg/cnvlib/commands.py", line 758, in do_scatter

  File "build/bdist.linux-x86_64/egg/cnvlib/plots.py", line 367, in gene_coords_by_name
ValueError: No targeted gene named 'AJUBA' found

Any ideas?

[etal]

I think your BED file has no rows labeled "AJUBA" by itself, only "AJUBA;RP11-298I3.5" and perhaps other joined names. So, while I look into a general fix, you should be able to do:

 cnvkit.py scatter Sample.cnr -g AJUBA;RP11-298I3.5

Or even:

 cnvkit.py scatter Sample.cnr -g AURA_12,AJUBA;RP11-298I3.5

Pre-processing the target BED file beforehand with "--short-names" would fix this problem for you, at the expense of losing some gene names and having to figure out separately where they are.

The general fix (not yet implemented) will probably involve creating an index of gene names split by , on the fly, and searching that separate data structure with your query. So ; would be used to hide from CNVkit the fact that these rows really represent 2 or more genes, and , (after the fix) would be used for this index.

[chapmanb]

Thanks Eric, the split gene name approach you mention is exactly what we'd be looking for. The general idea is we'd like to be able to label with multiple genes if a region overlaps (or is equal distance from) more than one. That way we don't have to come up with a prioritization approach for cases where we have multiple. We can swap back to using , for splitting genes later if that's the more correct way to separate what we want to be different genes. Thanks again for looking at this.

[mjafin]

Thanks @etal and @chapmanb, I think I was jumping the gun a bit here!

[etal]

I think the last commit provides the feature you're looking for. Could you try it and let me know?

[mjafin]

@etal many thanks, sounds great! I'll give it a go first thing tomorrow.

[mjafin]

Not sure if I'm doing something wrong, but tried what I think is the latest installation and am getting this:

Traceback (most recent call last):
  File "/group/ngs/src/bcbio-nextgen/0.8.8/rhel6-x64/anaconda/bin/cnvkit.py", line 9, in <module>
    args.func(args)
  File "build/bdist.linux-x86_64/egg/cnvlib/commands.py", line 732, in _cmd_scatter

  File "build/bdist.linux-x86_64/egg/cnvlib/commands.py", line 770, in do_scatter

  File "build/bdist.linux-x86_64/egg/cnvlib/plots.py", line 367, in gene_coords_by_name
ValueError: No targeted gene named 'MTND2P28' found

(is line 367 where this error should be coming from in the latest version?)

Here's a .cnr file line with MTND2P28 on it:

chr1    564297  565169  MTND1P23,MTND2P28,RP5-857K21.4,RP5-857K21.5     0.565725        1.0

[etal]

The error should be coming from line 369 now, not 367. Did you pull from
GitHub? This code wasn't in the recent 0.4.0 release.

On Fri, Apr 17, 2015 at 7:52 AM, Miika Ahdesmaki [email protected]
wrote:

Not sure if I'm doing something wrong, but tried what I think is the
latest installation and am getting this:

Traceback (most recent call last):
File "/group/ngs/src/bcbio-nextgen/0.8.8/rhel6-x64/anaconda/bin/cnvkit.py", line 9, in
args.func(args)
File "build/bdist.linux-x86_64/egg/cnvlib/commands.py", line 732, in _cmd_scatter

File "build/bdist.linux-x86_64/egg/cnvlib/commands.py", line 770, in do_scatter

File "build/bdist.linux-x86_64/egg/cnvlib/plots.py", line 367, in gene_coords_by_name
ValueError: No targeted gene named 'MTND2P28' found

(is line 367 where this error should be coming from in the latest version?)

Here's a .cnr file line with MTND2P28 on it:

chr1 564297 565169 MTND1P23,MTND2P28,RP5-857K21.4,RP5-857K21.5 0.565725 1.0

—
Reply to this email directly or view it on GitHub
#15 (comment).

[mjafin]

OK, code is now up to date and things work great. Many thanks again for cnvkit, it's a great tool!