tamsen
commented
5 years ago Hi Azita,
Thanks for your interest.
That really depends on the quality and settings of your aligner. If you have fairly clean alignments, its not really needed. But if the aligner tends to make a lot of classification errors between SNPs and indels at the end of reads, then doing indel realignment might help. In our testing, Pisces alone handles indels much better than GATK alone (due to Pisces' "indel collapsing" algorithm) , and so we don't always use an indel aligner in our workflows.
That being said, we recently released our own indel realinger, Gemini, which you are welcome to use: https://github.com/Illumina/Pisces/wiki/Suggested-Pipeline-Configuration-5.2.10
https://github.com/Illumina/Pisces/releases
https://github.com/Illumina/Pisces/wiki/Gemini-5.2.10-Design-Document
best Tamsen
Excerpt from PiscesBioinformaticsApplicationNote_SupplementaryMethods.docx:
"The highest quality results can be obtained by using an indel-realigned, stitched bam."
"The full Pisces Suite workflow is described in Figure 1, below. The Pisces Suite includes the Pisces Stitcher, Pisces Variant Caller, Pisces Variant Quality Recalibrator (VQR), and Pisces Variant Phaser (Scylla). Pisces Suite does not currently provide an indel realigner, although this is a work in progress. Currently, the indel realigner provided by GATK [25,26] may be used."
GATK IndelRealigner command has been removed. Is this step required in order to get the highest quality results?
Thanks, Azita