Stupid question about abyss

[cli]

Hi, all,

I have 100 bp pe reads from Hiseq for target genes with size around 200bp. I tried abyss to de novo assemble them.

I used k=40, n=10. When I compared the contigs and the reference sequence, which I used as positive control, I found that there were many small contigs overlapping each other by more than 50bp. Why abyss didn't assemble them into a longer contig? After I increase the k to 60, I saw longer contigs, but there were still overlapping but separate contigs. I am naive to the de bruijn graph and have no ideas about how abyss works. What can I do to get longer contigs instead of shot overlapping ones. Should I increase k more or change the n parameter.

Another question, when I used k=60, did I excluded all reads that are shorter than 60 bp from analysis?

Thanks in advance for your help.

[peromhc]

I am naive to the de bruijn graph and have no ideas about how abyss works.

Have you looked at the ABySS listserve (https://groups.google.com/forum/?hl=...um/abyss-users). A wealth of info there that you may find useful.

[Wallysb01]

I also highly recommend taking your questions to the ABySS user group. The developers are quite on top of answering questions.

However, I can tell you it would helpful to know what commands including all the options you used. It sounds to me like you might have just done the initial step (the ABYSS step which is the SE assembly only). That's fine to start with, but you should also continue down the assembly pipeline using abyss-pe. Many of the next steps will merge these overlapping contigs.