Details on Pacbio SNP calling using bbmap callvariants.sh.
Using non-corrected reads, I could call SNPs using these parameters on 4 human mito sample with coverage of 2000+
callvariants.sh in=in.bam out=out_SNV.vcf ref=/lager2/rcug/seqres/HS/mito.fa minquality=1.0 minqualitymax=2 overwrite=t ploidy=1 minscore=2.0 minpairingrate=0 usepairing=f useidentity=f
Note the score parameters are set exceedingly low. I did have them higher, but continually reduced them to get any results at all.
BBMap VCF Qual score distributions of 2-6 were output.
I called SNPs again with the same parameters on reads that had been corrected by Canu with default settings, genome size set to 16.5k. Corrected coverage was low, around 30 (previously around 2000+). A surprising reduction.
BBMap VCF Qual score distributions of 26-42 were output.
Thanks for the nice VCF formatted pacbio SNPs Brian, now I can process downstream with snpeff etc. This wasn't the case with samtools, Freebayes or SMRT analysis.