short_long does not support gtf without gene line.

[dputhier]

It seems that short_long does not support files without gene lines. I think the gene_ids should could taken from the transcript lines or even the exon line since it is very common to have no gene line. What do you think about this ?

 # Here toto.gtf has no gene line
gtftk convert_ensembl  -i  toto.gtf | gtftk  select_by_key -k feature -v gene -n | gtftk short_long
# segfault
gtftk convert_ensembl  -i  toto.gtf |  gtftk short_long
# ok

[fafa13]

It is done exactly like you want. gene_ids are taken from transcript line if such a line exists. If not, they are taken from exon lines.

[fafa13]

My previous comment is for convert_to_ensembl function. You're right, get_shortest/longest_transcript functions don't work if there is no gene lines. I will try to fix that ...

[fafa13]

I just pushed a fixed version for this bug. Can you test it and tell me if it works ?

[dputhier]

This does not seem to fix the issue regarding #42.
The command ends up with a segfault. Sure you have pushed ?

  gtftk get_example -d mini_real |  gtftk select_by_key -V 2 -C -k feature -v exon  | gtftk short_long
  Segmentation fault: 11

[dputhier]

This one is not fixed

    $ gtftk get_example -d mini_real |  gtftk select_by_key -V 2 -C -k feature -v exon  | gtftk short_long
        |--- 11:13-DEBUG-select_by_key : Creating a GTF instance.
        |--- 11:13-DEBUG_MEM-select_by_key : GTF created (#lines=137670, file=-, ptr_addr=0x7fea45b21b20, id=4525751952, nb=1).
        |--- 11:13-DEBUG-select_by_key : Calling extract_data.
        |--- 11:13-DEBUG-select_by_key : Calling select_by_key (key=feature, value=exon)
        |--- 11:13-DEBUG-select_by_key : Creating a GTF instance.
        |--- 11:13-DEBUG_MEM-select_by_key : GTF created (#lines=64251, file=-, ptr_addr=0x7fea48198370, id=4525482768, nb=2).
        |--- 11:13-DEBUG_MEM-select_by_key : GTF deleted (#lines=137670, file=-, ptr_addr=0x7fea45b21b20, id=4525751952, nb=1).
        |--- 11:13-DEBUG-select_by_key : Writing a GTF (#lines=64251, file=-, ptr_addr=0x7fea48198370, id=4525482768, nb=2).
        |--- 11:13-INFO-select_by_key : GTF written.
        |--- 11:13-DEBUG_MEM-select_by_key : GTF deleted (#lines=64251, file=-, ptr_addr=0x7fea48198370, id=4525482768, nb=2).
    Segmentation fault: 11

[fafa13]

If you add a "gtftk convert_ensembl" after select_by_key, there is no more error. I will try to do what you suggested before, that is to get transcript_ids from exon lines.

[fafa13]

I have made a push to try to fix this one. The changes are big so I made a copy of the function select_transcript that I called select_transcript2. Can you test your example in python after replacing the calls to select_transcript by the the new function ?

[dputhier]

I will try this asap.

[dputhier]

I end up with 61 failures out of 414 tests. The select_transcript2 function also seem to also have problems:

  gtftk short_long -i gtftk/data/simple_03/simple_short_long.gtf
  >segfault

I have created a branch in gtftk containing your last libgtftk version (feature_libgtftk_bc072e5). Maybe it will help to debug.

[fafa13]

I can't see the select_transcript2 function in the new branch. Is it normal ?

[dputhier]

did you try a:

git pull
git checkout feature_libgtftk_bc072e5

[fafa13]

I did even more :
git clone [email protected]:dputhier/gtftk.git
git checkout feature_libgtftk_bc072e5

And still no select_transcript2 function.

[fafa13]

This one seems to be fixed by the last push.

[dputhier]

To be tested.