How do I interpret my 'bcftools +fixref' output and troubleshoot it.

Question

[sandKings]

Using GATK method to call variants from RNASeq data, I have VCF files from 30 patients. There's a lot of missing genotype data and I need to impute missing information from Sanger imputation service. When I uploaded my vcf file to Sanger, I received this error:

--- Aborted Job --- The input file sanity check failed, "bcftools norm -ce" exited with the following message: Reference allele mismatch at MT:150 .. REF_SEQ:'C' vs VCF:'T'

The error report further recommended that I try ref allele mismatch correction using bcftools.

Based on documentation on this page https://samtools.github.io/bcftools/howtos/plugin.fixref.html I tried a ref allele mismatch correction using : 'bcftools +fixref report.vcf -- -f ref.fa'

My output looks like this:

# SC, guessed strand convention
SC  TOP-compatible  0
SC  BOT-compatible  0
# ST, substitution types
A>C 1058    3.3%
ST  A>G 5927    18.5%
ST  A>T 726 2.3%
ST  C>A 1069    3.3%
ST  C>G 1512    4.7%
ST  C>T 5513    17.3%
ST  G>A 5551    17.4%
ST  G>C 1509    4.7%
ST  G>T 1107    3.5%
ST  T>A 744 2.3%
ST  T>C 6174    19.3%
ST  T>G 1064    3.3%
# NS, Number of sites:
NS  total           33609
NS  ref match       31954   100.0%
NS  ref mismatch    0   0.0%
NS  skipped         1655
NS  non-ACGT        0
NS  non-SNP         1647
NS  non-biallelic   8

In the vcf file itself, I manually found chrM 150 and it shows:

#CHROM  POS ID  REF ALT QUAL    FILTER  INFO
chrM    150 .   T   C   1536.77 SnpCluster  AC=2;AF=1.00;AN=2;DP=29;ExcessHet=3.0103;FS=0.000;MLEAC=2;MLEAF=1.00;MQ=60.00;QD=34.76;SOR=1.460

Could anyone please help me troubleshoot this step. Thanks!

[Pierre Lindenbaum]

are you using the very same REFerence genome that was used to map he reads ? MT sequences are not always the same.

[sandKings]

Hi Pierre! Thanks so much for responding. Yes, in fact, to avoid the liftover procedure, I did the variant calling from scratch using GRCh37 since Sanger Imputation Service works on GRCh37.

In my very first impute attempt on Sanger, I got an error: faidx_fetch_seq failed at chr1:567242 and the error report further suggested that I change the UCSC-style chromosome names to Ensembl-style chromosome names which I did using:

bcftools annotate -Oz --rename-chrs ucsc2ensembl.txt ucsc.vcf.gz > ensembl.vcf.gz

as described on this page

EDIT: I'm wondering if it's relevant to point out that the variant calling was done on RNASeq data via GATk and the RNASeq is a 50bp SR dataset.