train.f.gb empty

[diriano]

My train.f.gb file was empty. But I had plenty of good genes from RNASeq data. It turned out a problem in Augustus-3.3.3/scripts/filterGenesIn_mRNAname.pl that was not parsing correctly the transcript names.

I changed the line

if ( $_ =~ m/transcript_id "(.*)";/ ) {

for this one

if ( $_ =~ m/transcript_id "([0-9_t]*)"/ ) {

And now I have genes in train.f.gb

[phyden]

Hi,

I observed the same issue. This also matches the observations in #125
The reason is apparently, that GeneMark-ES version 4.48 adds some more attributes after the transcript_id such as cds_type and count, which it did not in version 4.47. This is due to the very unspecific wildcard in the regex .*. It expands to everything between two double quotes, and therefore the extracted transcript_id string will not match the gene_id in the genbank file anymore.

Using an older version of GeneMark-ES would also be a solution, but there is no source where they can be obtained (at least I could not find any) and it's never a good idea to rely on older software unless there is a very good reason for this.

br,
Patrick

edit: I am not very experienced, with perl and augustus, but it might be, that the regex @diriano proposed is not working for all applications of augustus. I would therefore recommend to use the same regex but in a non-greedy variant, which seems to work as well for the needs here:

# instead of
f ( $_ =~ m/transcript_id "(.*)";/ ) {

#simply adding a questionmark (?)
f ( $_ =~ m/transcript_id "(.*?)";/ ) {

edit2: I saw that this was already fixed in a different way in the master branch of Augustus after the last release. With the next releases this issue will probably not recur.

[KatharinaHoff]

I hope so.