Methylation base provided by the BED file is inconsistent with the genomic base

[shizhuoxing]

I performed pileup using modkit v0.5.0_5120ef7 to generate a BED file and manually examined a subset of sites. It was found that the reported m6A sites are not actually "A" bases in the genome. For example:

Row in the BED file:
Seq1 5 6 a 10 + 5 6 255,0,0 10 40.00 4 4 2 0 3605 4311 0

Genomic sequence (Pos 1→70):
GCAGCGCAAGTAGGTCTACAAGACGCTACTTCCCCTATCATAGAAGAGCTTATCACCTTTCATGATCACG

Does modkit, during pileup, only check for m6A tags on reads? If there is a shift in the alignment of a site on a read due to mismatches or indels, does modkit not verify it against the reference genome?

[ArtRand]

Hello @shizhuoxing,

Modkit <0.6.z pileup would report records that have mismatches to the reference base like this. For example, if the reference base is 'A' but there is a read with a C>A mismatch any base modifications on C bases will be reported. If you update to v0.6.1 and use --modified-bases the output table will only base modification records for the reference primary sequence base. There is additional documentation on migrating to v0.6.0+.