vcf_combo.py assumes INFO AF=1 or AF=2

[Redmar-van-den-Berg]

However, AF contains a count for each ALT allele that is present. In cases where there is more than one ALT allele, there will also be multiple values in the AF field.

This happens in quite a few samples, mainly in cases of indels:

VAL201_S1_no_homref.vcf	WARNING:root:work:101 - chr22:42132026 has AC=(1, 1), skipping variant completely
VAL204_S4_no_homref.vcf	WARNING:root:work:101 - chr22:42132024 has AC=(1, 1), skipping variant completely
VAL205_S5_no_homref.vcf	WARNING:root:work:101 - chr22:42130044 has AC=(0, 1), skipping variant completely
VAL206_S6_no_homref.vcf	WARNING:root:work:101 - chr22:42132026 has AC=(1, 1), skipping variant completely
VAL210_S10_no_homref.vcf	WARNING:root:work:101 - chr22:42132024 has AC=(1, 1), skipping variant completely
VAL211_S11_no_homref.vcf	WARNING:root:work:101 - chr22:42132025 has AC=(1, 1), skipping variant completely
VAL213_S13_no_homref.vcf	WARNING:root:work:101 - chr22:42132024 has AC=(1, 1), skipping variant completely
VAL214_S14_no_homref.vcf	WARNING:root:work:101 - chr22:42132027 has AC=(1, 1), skipping variant completely
VAL215_S15_no_homref.vcf	WARNING:root:work:101 - chr22:42132024 has AC=(1, 1), skipping variant completely
VAL216_S16_no_homref.vcf	WARNING:root:work:101 - chr22:42132024 has AC=(1, 1), skipping variant completely

[marksantcroos]

Btw, should I interpret this as s/AF/AC/g?

[Redmar-van-den-Berg]

Yes, the script checks for the AC field. However, the same logic applies to the AF and the AC field, i.e. they can contain multiple values when there are multiple ALT alleles.