Wrong attributes after converting from BAM

[olekto]

Dear Luca,
I have mapped some IsoSeq3 transcripts with minimap 2, and have converted the BAM file to GTF with this script:
https://github.com/EI-CoreBioinformatics/mikado/blob/master/util/bam2gtf.py

When running mikado prepare, I get a lot of these errors:
Wrong attributes (['"']) in line: LG01 bam2gtf transcript 673836 675790 60 - . gene_id "transcript/26892.gene"; transcript_id "transcript/26892"; identity "NA"; coverage "90.56"; cigar "2S100M2D202M1I333M13I512M6D1M1D444N354M156S"; NM "31"; ms "1445"; AS "1409"; nn "0"; ts "-"; tp "P"; cm "444"; s1 "1428"; s2 "1162"; de "0.00860000029206276"; SA "GmG20150304_scaffold_6248,682,-,1516S157M1576D1S,0,4;";
Can you understand what is happening? Are these transcripts lost?

Thank you.

Ole

[lucventurini]

Dear @Tierhon,
Thank you for reporting this issue. I think I might have fixed it already today, as I was working specifically on this part of the code (see e41122d). We might be able to verify the fix in two ways:

• if the data is not confidential, you could send me a small subsection of the Bam, so that I can check directly
• alternatively, you could install the software directly from the github repo and check whether the changes from today solve your issue.

Hope this helps.

Kind regards

Luca Venturini

[olekto]

Dear Luca.

I am feeling really stupid right now. I cloned the git repository, and installed it by
python3 setup.py test; python3 setup.py bdist_wheel; pip install dist/*whl

This was done with in a miniconda3 environment. But, where is the binary? The mikado file that drives the procedures. I can't find it anywhere, and there does not seem to be any apparent issues in the installation.

Ole

[lucventurini]

Hi Ole,
the binary should have been installed in your PATH - ie inside the Miniconda3 folder.
You should be able to test this by doing:

$ python3
[1] import Mikado
[2] print(Mikado.__version__)
1.5

Best
Luca

[olekto]

Hmm, something went wrong somewhere. I am sorry. I get this result:

Python 3.6.7 | packaged by conda-forge | (default, Feb 20 2019, 02:51:38) 
[GCC 7.3.0] on linux
Type "help", "copyright", "credits" or "license" for more information.
>>> import Mikado
>>> print(Mikado.__version__)
Traceback (most recent call last):
  File "<stdin>", line 1, in <module>
AttributeError: module 'Mikado' has no attribute '__version__'

I can send you the python setup.py test result on email. Not sure how to troubleshoot this.

Ole

[lucventurini]

From the mail thread: the problem is that there are some attributes in the GTF which contain, wrongly, the character ";" within. This confuses the parser and causes the problem.
To be fixed ASAP.

[olekto]

Great! Thank you.

[lucventurini]

Dear @Tierhon , the problem should now be fixed. Please let me know!

[olekto]

I had no issues loading these files now. Thank you!

[lucventurini]

Excellent! And thanks to this issue, I should also have fixed the problems with python setup.py test. So double progress :-)