GATK Haplotype caller missing calls

[nvteja]

Hello friends,

I am using GATK Haplotype caller to call snps and indels from a GATK realigned, re-calibrated bam file (following GATK best practices). The only options I have changed from default values are -stand_call_conf and -stand_emit_conf, both set to 20. When I was comparing the GATK vcf file to samtools snp file for concordance I have noticed that GATK Haplotype caller missed some snps which look very real in IGV. Here is pileup for one of those snp positions from samtools (hg19)

8 6338306 C 140 ..A...A...A.a,...AAAA.,aA,.A...,,A,.A,a.aa,aaa.,.A,,A..,,AA.,.AAAAAa..,a,a,,,AAA.aA,aaa,,.AAaaaAAa,aa..,.a,a.Aa,.AA..,,aa,,,,A,aa,aaa,a,a,,a 000111111133;;55537:79;:<;<<;<<<<=<;=:<<::;;=<<<>:<<8<<<<<8<<<:=8<<;<<<==;<;==<<<=<;===<<<<=<<<==;;6:<<9<969<<77;<<<;321201./<-.0-/.0.0/./..

Also, samtools view shows that consensus base is "A"

#CHROM POS ID REF ALT QUAL FILTER INFO FORMAT 11.0422
8 6338306 . C A 168 . DP=140;VDB=0.0241;AF1=0.5;AC1=1;DP4=36,36,32,34;MQ=51;FQ=170;PV4=0.87,1,1,1 GT:PL:GQ 0/1:198,0,203:99

I have tried running GATK with both call conf and emit conf to 10 and still couldnt see this variant. I do not want to reduce these to 0 as it will result in lot of FP calls. Am I missing anything? Any help with this is very appreciated!

Thanks a lot!
Teja.