mpileup to STRUCTURE

[jgibbons1]

Hi all,
Thought I would see if anyone has a different (better!) pipeline for collecting a set of SNPs from many bam files for use with STRUCTURE or other population genetic software. The end goal is to have a matrix with SNPs as rows and samples in column.

Ex.
x y z
Site 1 T T G
Site 2 G A C
Site 3 A G G

1. Create mpileup file
2. Call SNPs (with your favorite software) for each sample
3. Create merged SNP list, with the end of goal of having each position where there is a SNP in at least one sample
4. Create consensus fasta file from mpileup for each sample
5. Extract consensus nucleotide from each position (#3) from consensus fasta (#4)
6. Merge files

One problem is that some SNP sites found in particular samples may not pass quality filters in other samples. I would love to hear of other pipelines.

Thanks!

[SNPsaurus]

Have you tried just sending the mpileup to bamtools call?
samtools mpileup -gu -t DP -f ref.fasta -b bam_file_list.txt| bcftools call -cv - > genotypes.vcf
You'll need to sort the bam files, with samtools sort for instance

or the equivalent from freebayes
freebayes -f $ref_fasta_file bam_string > vcf_name
You'll need to "read group" the bams with bamaddrg -b bamfile and then index with bamtools index.

Either of these should be filtered with vcftools or vcflib to keep the SNPs with a certain presence in the population, using calls of a certain depth for each sample, etc.

[jgibbons1]

Thanks for these suggestions. I actually haven't explored bcftools or freebayes so should have a busy few days

I will post an update once I come to a satisfactory solution.