Indel detection in high coverage amplicons sequenced by MiSeq

[lpalacios]

Hello everybody,
I have started working with MiSeq and I am trying to sequence custom amplicons with 500 cycles kit (250 nucleotides sequenced by R1 and R2). I have not managed to detect all the indels, specially long indels (above 18-40 nucleotides). My library contains 48 amplicons of 300 nucleotides on average and they have high coverage (around 10000 reads each). As the MiSeq reporter results are not accurate, I am using bwa to generate .bam, then samtools to sort and index the .bam files and finally GATK to detect variants. I have tried with -glm BOTH, and also witn -glm INDEL but both failed.
When I load the .bam in the IGV I can see that there is an insertion or a deletion in the corresponding nucleotide and all the SNPs are correctly shown but then when I obtained the .vcf by GATK I obtain a lot of false positives and false negatives and I have no long indels.
I have also tried PINDEL but I have not detected long indels either, perhaps because the R1 and R2 sequences are overlapped.
Could anybody tell me how can I improve my variants detection, specially the indels?

Thanks in advanced!

Lourdes

[rlong]

For pindel, you might improve results by running samtools fixmate on your bam. Pindel will recover some unmapped read mates, but only if they are flagged as such, and located alongside the mapped mate. So if you had stringent mapping quality restrictions during alignment, you might find more support for larger indels this way.

You could also use samtools to call variants and see what sort of results you get there.

[alexholman]

This post is good and stale, but I'll update my solution for posterity.

Try the package freebayes "Bayesian haplotype-based polymorphism discovery and genotyping"

https://github.com/ekg/freebayes

The caller seems to work well on amplicon data and ended up being the cleanest and most complete VCF file (with ref and alt allele frequencies).