search

ababaian / serratus

Ultra-deep search for novel viruses
http://serratus.io
GNU General Public License v3.0
250 stars 32 forks source link

DARTH: Uninitialized value error from VADR #205

Closed taltman closed 3 years ago

taltman commented 3 years ago 
2020-07-16T15:13:40.662-07:00

# sequence file: /serratus-data/AY394999.1.fa

2020-07-16T15:13:40.662-07:00

# output directory: /serratus-data/AY394999.darth/AY394999

2020-07-16T15:13:40.662-07:00

# force directory overwrite: yes [-f]

2020-07-16T15:13:40.662-07:00

# .cm, .minfo, blastn .fa files in $VADRMODELDIR start with key <s>, not 'vadr': corona [--mkey]

2020-07-16T15:13:40.662-07:00

# model files are in directory <s>, not in $VADRMODELDIR: /darth/vadr-models-corona-1.1-1 [--mdir]

2020-07-16T15:13:40.662-07:00

# set max allowed memory for cmalign to <n> Mb: 64000 [--mxsize]

2020-07-16T15:13:40.662-07:00

# - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -

2020-07-16T15:13:53.225-07:00

# Validating input ... done. [ 12.6 seconds]

2020-07-16T15:17:51.213-07:00

# Classifying sequences (1 seq) ... done. [ 238.0 seconds]

2020-07-16T15:18:40.031-07:00

# Determining sequence coverage (NC_004718: 1 seq) ... done. [ 48.8 seconds]

2020-07-16T15:21:34.850-07:00

# Aligning sequences (NC_004718: 1 seq) ... done. [ 174.8 seconds]

2020-07-16T15:21:34.973-07:00

Use of uninitialized value $uapos in concatenation (.) or string at /vadr/vadr/v-annotate.pl line 3757.
taltman commented 3 years ago

Confirmed this bug. Reported upstream to Eric at NCBI: https://github.com/nawrockie/vadr/issues/21

rcedgar commented 3 years ago

@tomer this bug is independent of the PFAM annotations, right? Is it possible to run the PFAM alignments on the GB entries that crashed outside of VADR while we're waiting for the bug to be resolved?

taltman commented 3 years ago

It is not independent, as the Pfam annotations are currently used to identify the maturation proteins on ORF 1ab, and the gene-calling of ORF 1ab comes from VADR.

The first thing I'd like to see is for @rchikhi to re-run these few samples, and see if that clears up some of the failed runs. I can then look at the output from that second run of each sample, to see what the issue is.

taltman commented 3 years ago

As I suspected, merely re-running them seemed to clear up ~96% of the issues. The two accessions that caused that obscure VADR crash have been reported up-stream to Eric.

taltman commented 3 years ago

Thanks @rchikhi for re-running. @rcedgar , please leave this open as a reminder on my part to follow up with Eric.

taltman commented 3 years ago

FYI, I filed this bug in the VADR repo, and Eric responded that he's currently working on a fix:

https://github.com/nawrockie/vadr/issues/21

rchikhi commented 3 years ago

looks like it's been temp-fixed https://github.com/nawrockie/vadr/issues/21#issuecomment-661279793

taltman commented 3 years ago

Yep, he provided me with a work-around for now, and a proper fix will go out in the next VADR release, which will be a few weeks from now. I'll incorporate it into the updated version of DARTH that I'll send to you this evening.

taltman commented 3 years ago

Confirmed that my implementation of the work-around works for the example above. Closing!