Generating simulated paired end reads

[whaleberg]

I'm writing a NSG simulator as part of a project for my school. I'm trying to simulate fastq paired-end data, but I'm unsure of the formatting for doing so. I'm producing simulated reads and successfully aligning them in BWA for single ended data, but when I try to produce pairs I end up with various alignment errors. I read that for BWA, the reads in the second alignment file should be the reverse complement.

If I have a base sequence and a simulated a pair of reads from each end like this...

            AAAGGGTTCTC
read        AAAG
read               TCTC

I'm outputing
file1
read/1 AAAG

file2
read/2 TCTC

(along with the quality scores and the rest of the formatting)

Is that the way they should be? It doesn't seem to work, so I'm guessing not, but neither does not doing the reverse complement. I suspect I'm missing something about how pairs of reads are represented.


I may also have the naming conventions wrong. Should paired reads be separated into 2 different files and labeled <read-name>/1 and <read-name>/2? Ultimately they get rolled into 1 file, so should I be putting them together into 1?

I'm not sure if I have a software bug and am just producing wrong data, or if I'm producing the right thing but formatting it wrong.

Does anyone have an example of correct formatting that I could use as a template? I have had trouble locating an example.

Any help would be appreciated.

[proteomania]

If you are talking about the illumina standard pair-end seq, i think i should be

file1
read/1 AAAG

file2
read/2 GAGA

Our illumina data look this way. The mate are on opposite strands although the library is not strand specific.

hope it helps.

[whaleberg]

Thank you, that is helpful. I had tried doing it that way, but I was still getting weird results. It must be a bug in something else, not that way I'm formatting them.

[husamia]

do you mean that they are not aligning correctly or what?