Missing CheckV output when there are multiple prophages in contig about emg-viral-pipeline

[hoelzer]

Hey @Ales-ibt - wow, thanks for the detailed report and visuals!

[Ales-ibt]

Hello again 😄

After updating the Virify version we are locally running, now I'm getting the output of checkV and the gff table. However, the program is reporting one single prophage when I expect multiple. Also, the coordinates of the prediction on the gff file don't correspond to the location of any prophage in the prophages.fna file. An example here:

$ cat prophages_original.fasta | grep '>'

b5_04_1 prophage-181138:243992
b5_04_1 prophage-1725901:1775585
b5_04_1 prophage-2873420:2950084
b5_04_1 prophage-909689:958923
b5_04_1 prophage-690730:746532
b5_04_1 prophage-1508929:1577763
b5_04_1 prophage-3443947:3467484
b5_04_1 prophage-2728160:2758576

$ cat 08-final/gff/b5_04_mod_virify_contig_viewer_metadata.tsv
sequence_id contig virify_taxonomy start_of_first_viphog end_of_last_viphog checkv_provirus checkv_quality miuvig_quality
b5_04_1-start-6662-end-63413 b5_04_1 Caudovirales;;; 6662 63413 Yes High-quality High-quality

My guess is that CheckV is running on the whole contig and it is defining new coordinates for one single prophage prediction, instead of running on every predicted prophage.

I hope this info is useful to improve Virify output.

Bests,

Alejandra.

[mberacochea]

Perfect, thank you Ale. I'll take a look

[Ales-ibt]

Me again...

After taking a look at the output files and tracing back the checkV results, I have the following summary of problems associated with the integration of the outputs:

Problems related to the integration of predicted phages and quality:

1. CheckV is running on the full contig for prophages
2. Multiple phages occurring in the same contig are missing
3. Prophages coordinates in the gff don't correspond to coordinates prediction
4. Prophages with no CDS matching ViPhOG are excluded in the gff (is this for quality control?)
5. Viral genomes in full contigs are missing in the gff output

Problems related to the GFF file format (which is not validated):

1. Protein IDs in gff file are missing
2. Add the mobile_genetic_element feature and the corresponding coordinates. Then link the corresponding CDS using the 'Parent' key in the attributes field
3. Fix the CDS coordinates in the context of the full contig instead of the subsequence

Find attached some figures I prepared for this report.

Bests,

Alejandra.
virify_bug_Feb2023.pdf

[mberacochea]

Fixed by #95 and #94