weird output with BWA paired end alignment

[jmt]

Hi all,

I'm running into something strange.

I'm trying to do paired-end alignment with BWA.

I did the following:

bwa aln fasta pair1.fastq > pair1.sai
bwa aln fasta pair2.fastq > pair2.sai

Then, I did

bwa sampe fasta pair1.sai pair2.sai pair1.fasta pair2.fastq > bam

However, when I looked at my bam, I got alignments such as this:

seq1 117 chr10 49643024 0 * = 49643024 0 CCCGTATAACTCCATTCTTGGTTCTCTCCTCCAAGTCCATAACTTCCTTATATATCAATTCTTTCCACT
seq1 153 chr10 49643024 37 100M = 49643024 0 AGACAGCAGTCTAGAAGCAGCAGCTGGGCCTCTGGGCTGCTGTAGATGACTACTTGGGCCATCGGGGGG

My question is: shouldn't seq1 (first in pair; flag 117) have a "*" under column 3, signifying that it's unaligned?

What is going on here?

[sudders]

We've noticed this as well. BWA seems to give a position to unaligned reads.

[jp.]

How about if give commands like this:
bwa aln -t 4 -f Sample_R1.sai ./hg19.fasta ./Sample_R1.fastq
bwa aln -t 4 -f Sample_R2.sai ./hg19.fasta ./Sample_R2.fastq