SEQanswers (
-   Bioinformatics (
-   -   Samtools on Chromosomes Versus Genome (

ramirob 04-10-2012 04:17 PM

Samtools on Chromosomes Versus Genome
Hi everyone,

Does anybody have experience running samtools on CASAVA alignments?

I am trying to call SNPs using samtools on a CASAVA/illumina alignment.

CASAVA outputs an alignment of the type name_export.txt.gz

and it automatically converts to BAM format when using their SNP caller.

Now I want to call SNPs using Samtools.

The alignment was done using each chromosome individually:


However, the BAM alignment provided by CASAVA is for the whole genome, so then should I use the whole reference genome?:

samtools mpileup -f ../genomes/Homo_sapiens/UCSC/hg18/Sequence/WholeGenomeFasta/genome.fa sorted.bam > sorted.mpileupOutput

the problem is that, in that case the mpileup output doesn't seem to recognize it as I get:

bash-3.2$ head -n 100 sorted.mpileupOutput
chr1.fa 3309 N 1 ^NT C
chr1.fa 3310 N 1 G C
chr1.fa 3311 N 1 C F
chr1.fa 3312 N 1 C F
chr1.fa 3313 N 1 C F

Any ideas??

Thanks in advance,


sdriscoll 04-10-2012 04:33 PM

try this. mpileup doesn't do too much. bcftools is the program that actually does the SNP calling.


samtools mpileup -uf ../genomes/Homo_sapiens/UCSC/hg18/Sequence/WholeGenomeFasta/genome.fa sorted.bam | bcftools view -bvcg - > sorted.bcf
bcftools view sorted.bcf > sorted.vcf
head sorted.vcf

bioinfosm 04-10-2012 06:38 PM

or you can try splitting the bam file by chromosome and then calling SNP on them separately.. you could also try starting from the fastq or qseq files, run alignment using something like bwa and then call variants

sdriscoll 04-10-2012 08:11 PM

indeed...i align data using BWA when i'm looking for SNPs.

AllanLindh 04-11-2012 05:22 PM

How to combine .vcf files into whole genome
This discussion is very useful, but when I run mpileup plus bcftools to produce .vcf files for individual Chrm, I have found no way to combine them into whole genome .vcf file for further processing. vcf-concat seems like it might do it, but it never runs due to diverse errors. Most recent is:
The use of -1 for unknown number of values is deprecated, please use '.' instead.
FORMAT=<ID=PL,Number=-1,Type=Integer,Description="List of Phred-scaled genotype likelihoods, number of values is (#ALT+1)*(#ALT+2)/2">
I am running most recent version of samtools (0.1.18) -- has anyone else noticed a mismatch in the FORMAT specs between samtools and vcftools??

Any suggestions at all would be very welcom, this is driving me nuts.

Allan Lindh:confused:

ramirob 04-13-2012 11:01 AM


It seems that mpileup looking like:

chr1.fa 3309 N 1 ^NT C
chr1.fa 3310 N 1 G C
chr1.fa 3311 N 1 C F
chr1.fa 3312 N 1 C F

is a bad sign, it looks to me that it is not mapping to the ref. genome properly. I did as you suggested and everything was a SNP.

swbarnes2 04-13-2012 11:14 AM

Double check that the chromosome names in your .bam file match the chromosome names in your fasta. And I'd remake the fa.fai with samtools faidx, see if that completes without an error.

frc1230 05-19-2012 02:19 PM

I am currently trying to deal with exactly this problem: CASAVA alignments that choke mpileup. We use mpileup output directly for analysis with our own SNP finders, and it works very well with output from a BWA-based aligner (stampy). But with CASAVA output, as stated in another response, mpileup thinks everything is a SNP, all reads with phred scores of 0. This is not the case, fortunately, as the same read data run through stampy produces an alignment file that behaves well with mpileup.

ramirob 05-22-2012 03:38 AM

I must admit I just gave up on Casava alignments, just use BWA and I am exploring other aligners.

frc1230 05-22-2012 08:19 AM

I like stampy-BWA, but it does take a lot of cpu time

All times are GMT -8. The time now is 08:07 AM.

Powered by vBulletin® Version 3.8.9
Copyright ©2000 - 2020, vBulletin Solutions, Inc.