Bowtie does not align my both ends as a pair

[Fad2012]

Dear All

I am facing a problem with bowtie. I generated a library using MetaSim software for two viruses (Serotypes). I used an empirical error model, and --><-- as read pairs layout, the number of reads is 5000.

I am trying to paired-end align this data in Bowtie using the following command:

bowtie -S ebwt_file -f -1 reads_1.fasta -2 reads_2.fasta file .sam

The problem is that Bowtie is giving very poor results when the command above would executed:

# reads with at least one reported alignment: 16 (0.32%)
# reads that failed to align: 4984 (99.68%)

While i am expecting to see the opposite figure. However; both ends are aligned perfectly when i ask Bowtie to align them individually.

I need to align both ends as a pair using Bowtie, i can't work out what might be the reason behind these poor results. I tried to find out a pattern by examining the pair-algined reads ( 0.32% which is 16 reads), but i failed to find any interesting figures.

Many thanks

[kmcarr]

What insert size did you simulate for your paired reads? bowtie's default for a maximum valid insert (-X/--maxins) is 250 bp. If the read pair does not meet that criteria bowtie will not call it a valid alignment.

[Fad2012]

I am using 330 as mean insert size, i will try to scale it down and give it a try! I am wondering then, bowtie wont be the best choice to align long inserts. What would you suggest?

thanks very much kmcarr.

[kmcarr]

You don't have to scale your insert size down as bowtie can align reads from larger inserts, just enter appropriate values for your insert size with bowtie's minimum and maximum insert size parameters (-I and -X). There are a number of parameters which bowtie uses to determine what is and is not a valid alignment and these need to be tuned to your particular experiment.

[swbarnes2]

1) Try setting your min and max insert size options to a huge range
2) Try bwa, which will align each on individually, and will return them, no matter what the insert size is.