Hi there,
I've just aligned 5 whole canine genomes to the CanFam3 reference using BWA for subsequent processing using the GATK pipeline. The sequencing was 100 bp paired-end reads done on an Illumina HiSeq 2500. This is the BWA command I've used - pretty straightforward (version 0.7.10):
/path-to-bwa/bwa mem -M -R '@RG\tID:001\tSM:sample_name\tPL:illumina\tLB:sample_name_lib1\tPU:barcode_001' /path-to-reference-genome/canFam3.fa /path-to-r1/sample_r1.fastq /path-to-r2/sample_r2.fastq > /path-to-output/sample_aligned.sam
The job runs just fine on our cluster here, but I've noticed that when I process the files downstream using GATK, that in 4 of the 5 files about ~50% of the reads are not passing the GATK BadMate filter. In 1 of the 5 files there are nearly none that fail that filter.
Interestingly, in the 1 file that I'm not seeing this problem, the DNA was originally extracted from whole blood using a Qiagen kit; in the 4/5 where the mate pairs are not aligning correctly, they were extracted using phenol chloroform.
I wanted to see if anyone had any thoughts as to whether there was something wrong with the way I'm doing my alignment or is it likely library prep or DNA extraction? It seems weird that the same command would result in such a different output if the underlying files are the same. But maybe there's a better way to do the mapping? Should I try Bowtie2? BWA fits so nicely into the GATK pipeline, but I obviously don't want to lose 1/2 of my reads.
Also interestingly, when I look at the BAM files in the 4/5 dogs with the bad mate pairing, the data looks very "splotchy" with lots of peaks and valleys in the alignment; in the 1/5 dog where there were was good mate pairing, the coverage looks extremely consistent across the genome.
Thanks for any advice you might have for a newbie!
I've just aligned 5 whole canine genomes to the CanFam3 reference using BWA for subsequent processing using the GATK pipeline. The sequencing was 100 bp paired-end reads done on an Illumina HiSeq 2500. This is the BWA command I've used - pretty straightforward (version 0.7.10):
/path-to-bwa/bwa mem -M -R '@RG\tID:001\tSM:sample_name\tPL:illumina\tLB:sample_name_lib1\tPU:barcode_001' /path-to-reference-genome/canFam3.fa /path-to-r1/sample_r1.fastq /path-to-r2/sample_r2.fastq > /path-to-output/sample_aligned.sam
The job runs just fine on our cluster here, but I've noticed that when I process the files downstream using GATK, that in 4 of the 5 files about ~50% of the reads are not passing the GATK BadMate filter. In 1 of the 5 files there are nearly none that fail that filter.
Interestingly, in the 1 file that I'm not seeing this problem, the DNA was originally extracted from whole blood using a Qiagen kit; in the 4/5 where the mate pairs are not aligning correctly, they were extracted using phenol chloroform.
I wanted to see if anyone had any thoughts as to whether there was something wrong with the way I'm doing my alignment or is it likely library prep or DNA extraction? It seems weird that the same command would result in such a different output if the underlying files are the same. But maybe there's a better way to do the mapping? Should I try Bowtie2? BWA fits so nicely into the GATK pipeline, but I obviously don't want to lose 1/2 of my reads.
Also interestingly, when I look at the BAM files in the 4/5 dogs with the bad mate pairing, the data looks very "splotchy" with lots of peaks and valleys in the alignment; in the 1/5 dog where there were was good mate pairing, the coverage looks extremely consistent across the genome.
Thanks for any advice you might have for a newbie!