Im using SAMTOOLS, GATK UG and HC for SNP calling from target sequencing paired-end data. In contrast to all the others, UG produces massive amount of mutations near read starts/ends. Quality trimming actually is not ably to handle this problem completely since there is sometimes aligments errors near read ends. I didnt find any usefull information about root mean square distance to reads ends or something like this that GATK keeps in information fields in VCF. So im interesting if there is any opportunity to skip mutations falling near read start/end using GATK or how can i avoid this problem. And also is there any ready solution to calculate mean distance or root mean square distance to the nearest read end from single base position based on sam/bam file?
Thanks in advance