issue with annotating simple vcf only contains 8 columns of VCFv4.1

[ipstone]

Hello,

Thank you for your work on gvanno! I have gotten gvanno working on my centOS 7 box, and the example annotation ran well. However, when I am trying to annotate a simple vcf file like the following, I ran into the error (probably on every line of variant):

...
            ERROR: Line ...: Format is not a colon-separated list of alphanumeric strings.
            ERROR: Line ....: Format is not a colon-separated list of alphanumeric strings.
            ERROR: Line ....: Format is not a colon-separated list of alphanumeric strings.
            ERROR: Line ....: Format is not a colon-separated list of alphanumeric strings.

The VCF file looks like the following:

##fileformat=VCFv4.1
#CHROM  POS     ID      REF     ALT     QUAL    FILTER  INFO    FORMAT
1       2115900 .       T       C       .       PASS    AN=3646;AC=11   GT
1       2115911 .       C       G       .       PASS    AN=3646;AC=2    GT
1       2115912 .       G       A       .       PASS    AN=3646;AC=1    GT
1       2115999 .       C       T       .       PASS    AN=3646;AC=4    GT
1       2116124 .       C       G       .       PASS    AN=3646;AC=10,0 GT
....

What might be the cause of the error? Is there a way to format the vcf to get it properly annotated?

Thanks in advance!

-- ipstone

[sigven]

Hi @ipstone, Thanks for reaching out. If you look more closely at the excerpt of your VCF file above, it contains 9 columns and not 8. A quick suggestion is that you simply remove the 9th (FORMAT) column, since you do not have any sample (genotype) data present (this should occur in column 10 if you have a FORMAT column, and I suspect that is the reason it fails).

Try to see if that might give you success.

best, Sigve

[ipstone]

Thanks @sigven for picking out the format issue, it helps.

After removing the extra column, I noticed some other issue/s, as my input vcf file was originally a phased germline variants file, so some lines have multiple alleles info. With the vcf_validate option kept on, it would run into warning/error of

ERROR: Line 158011: INFO SVTYPE must be one of: BND, CNV, DEL, DUP, INS, INV. Found SVTYPE was 'MEI'.
           ERROR: Line 163444: INFO SVTYPE must be one of: BND, CNV, DEL, DUP, INS, INV. Found SVTYPE was 'MEI'.
           ERROR: Line 164819: INFO SVTYPE must be one of: BND, CNV, DEL, DUP, INS, INV. Found SVTYPE was 'MEI'.
           ERROR: Line 166354: INFO SVTYPE must be one of: BND, CNV, DEL, DUP, INS, INV. Found SVTYPE was 'MEI'.

If given --no_vcf_validate option, gvanno ran all the way, but gave an error when exporting the tsv file (I think):

2020-09-15 20:51:07 - gvanno-gene-annotate - INFO - Completed summary of functional annotations for 9345 variants on chromosome 16
2020-09-15 20:51:17 - gvanno-gene-annotate - INFO - Completed summary of functional annotations for 11197 variants on chromosome 17
2020-09-15 20:51:18 - gvanno-gene-annotate - INFO - Completed summary of functional annotations for 4889 variants on chromosome 18
Traceback (most recent call last):
  File "/gvanno/gvanno_summarise.py", line 125, in <module>
    if __name__=="__main__": __main__()
  File "/gvanno/gvanno_summarise.py", line 23, in __main__
    extend_vcf_annotations(args.vcf_file, args.gvanno_db_dir, args.lof_prediction)
  File "/gvanno/gvanno_summarise.py", line 91, in extend_vcf_annotations
    csq_record_results = annoutils.parse_vep_csq(rec, gvanno_xref, vep_csq_fields_map, logger, pick_only = True, csq_identifier = 'CSQ')
  File "/gvanno/lib/annoutils.py", line 705, in parse_vep_csq
    assign_cds_exon_intron_annotations(csq_record)
  File "/gvanno/lib/annoutils.py", line 454, in assign_cds_exon_intron_annotations
    exon_pos_info = csq_record['NearestExonJB'].split("+")
AttributeError: 'NoneType' object has no attribute 'split'

might it be an issue coming from :

gvanno-validate-input - WARNING - Multiallelic site detected:8    48317851        CGTGTGTGT       CGTGTGT,CG
TGT,CGT,CGTGTGTGTGT,C,CGTGTGTGTGTGT 

I have parsed the mutiple allelic form of VCF to one varianet per line, will test again it that helps.

Thanks again , I think it's making progress. There is already the '...gvanno_ready.vep.vcfanno.annotated.vcf' output file, I guess I could manually parse out the annotation if I still have trouble with the last step in python.

[ipstone]

Using the one variant per line (formatted input vcf), still run to similar error at line 125:

2020-09-15 21:51:31 - gvanno-gene-annotate - INFO - Completed summary of functional annotations for 11197 variants on chr[10/2828]
7                                                                                                                                 
2020-09-15 21:51:33 - gvanno-gene-annotate - INFO - Completed summary of functional annotations for 4889 variants on chromosome 18
Traceback (most recent call last):                                                                                                
  File "/gvanno/gvanno_summarise.py", line 125, in <module>                                                                       
    if __name__=="__main__": __main__()                                                                                           
  File "/gvanno/gvanno_summarise.py", line 23, in __main__                                                                        
    extend_vcf_annotations(args.vcf_file, args.gvanno_db_dir, args.lof_prediction)                                                
  File "/gvanno/gvanno_summarise.py", line 91, in extend_vcf_annotations                                                          
    csq_record_results = annoutils.parse_vep_csq(rec, gvanno_xref, vep_csq_fields_map, logger, pick_only = True, csq_identifier = 
'CSQ')                                                                                                                            
  File "/gvanno/lib/annoutils.py", line 705, in parse_vep_csq
    assign_cds_exon_intron_annotations(csq_record)
  File "/gvanno/lib/annoutils.py", line 454, in assign_cds_exon_intron_annotations
    exon_pos_info = csq_record['NearestExonJB'].split("+")
AttributeError: 'NoneType' object has no attribute 'split' 

This line linked to annoutils.py line 454 : https://github.com/sigven/gvanno/blob/7c6affd7ddc1badeb1e20a415af5729faacbfd0a/src/gvanno/lib/annoutils.py#L454 might it be a csq_record NoneType error? - perhaps adding a None check here?

[sigven]

Looks like you have hit the target there. If you are able to share your VCF with me (or some parts of it) along with your assembly and configuration file, I can make a more robust check when looking for Exon Junction information (i.e. check for None), and tests that it works for your case.

thanks, Sigve

[sigven]

Fixed