Hello,
I've created a bash script that automates pindel for our Illumina MiSeq samples. Initially we had a great output from Pindel and picked up FLT3, NPM1, and KIT insertions we were looking for. However, now we are seeing some discoradances compared to our genescan and capillary electrophoresis results. Pindel is missing calls and I am clueless as to what might be the reason why. Would running breakdancer and including it output in the pindel run help; however, I highly doubt that SNPs in the anchor or sequencing errors could be the cause of this issue since the reads in IGV look clean. Any suggestions? How about changing things here:
-d/--min_num_matched_bases only consider reads as evidence if they map with more than this number of bases to the reference (default 30)
-a/--additional_mismatch Pindel will only map part of a read to the reference genome if there are no other candidate positions with no more than the specified number of mismatches position.
The bigger this value, the more accurate but less sensitive. (default value 1)
-m/--min_perfect_match_around_BP at the point where the read is split into two, there should at least be this number of perfectly matching bases between read and reference (default value 3)
-e/--sequencing_error_rate the expected fraction of sequencing errors (default 0.05)
-u/--maximum_allowed_mismatch_rate only reads with fewer mismatches with the reference genome than this fraction will be considered (default 0.1)
What are others using here since Pindel strict in adhering to that perfect match or single mapped read and looking at the unmapped at the breakpoint for its calls?
I've created a bash script that automates pindel for our Illumina MiSeq samples. Initially we had a great output from Pindel and picked up FLT3, NPM1, and KIT insertions we were looking for. However, now we are seeing some discoradances compared to our genescan and capillary electrophoresis results. Pindel is missing calls and I am clueless as to what might be the reason why. Would running breakdancer and including it output in the pindel run help; however, I highly doubt that SNPs in the anchor or sequencing errors could be the cause of this issue since the reads in IGV look clean. Any suggestions? How about changing things here:
-d/--min_num_matched_bases only consider reads as evidence if they map with more than this number of bases to the reference (default 30)
-a/--additional_mismatch Pindel will only map part of a read to the reference genome if there are no other candidate positions with no more than the specified number of mismatches position.
The bigger this value, the more accurate but less sensitive. (default value 1)
-m/--min_perfect_match_around_BP at the point where the read is split into two, there should at least be this number of perfectly matching bases between read and reference (default value 3)
-e/--sequencing_error_rate the expected fraction of sequencing errors (default 0.05)
-u/--maximum_allowed_mismatch_rate only reads with fewer mismatches with the reference genome than this fraction will be considered (default 0.1)
What are others using here since Pindel strict in adhering to that perfect match or single mapped read and looking at the unmapped at the breakpoint for its calls?
Comment