Transcriptom assembly

[baohua100]

6 million 35bp solexa reads

I use ssake to assemble but many reads contain -AAAAAAAAA or -TTTTTTTTT

Shoud I first remove these reads then assemble?

[bioinfosm]

Yes, removing them is a big help to aligners/assemblers.

These are usually artifacts of the edge effect on lanes...

[myrna]

If you have paired reads, I would suggest trying Velvet for your assemblies. ABySS may also be useful for this, though no assembler is really meant for transcriptome assembly.

Ryan

[baohua100]

single end reads

Now I use edena2.1.1 to assembly

[glacerda]

edena and velvet are the best de novo assemblers, however they were not designed to transcriptome data.

Be aware of the coverage bias os solexa technology. Regions of high GC have different (I don't remember exactly if it is higher or lower) coverage.

This would cause problems if you would like to use the data to infer the expression levels of transcripts. In addition to this, you would have to normalize by transcript size (which almost never is known a priori, unless you have a complete sequenced genome, but in this case you would not be assembling transcripts de novo)