Does anyone know of any known issues with the indelrealigner? The GATK is calling 1000s of SNPs on one genome I have due to bad realignments. It appears the target finder identifies regions of the genome that are essentially perfectly aligned, but when the realigner gets to these areas it remits the reads as a new alignment that is a train-wreck compared to what the re-aligner started with. Am going to investigate further but thought I would check if this rings any bells. I am using the flag -model USE_READS and have no known indels to work with.