Trimming 5' end of RNAseq reads for de novo assembly

[Kennels]

Hi,

I know this has been discussed previously (http://seqanswers.com/forums/showthr...ptome+assembly) , but I'm wondering if anyone can share their experiences on de novo transcriptome assembly with trimming and not trimming the 5' end due to the 'random' priming bias of illumina RNAseq reads. How much could this affect the assembly really?

Can anyone point to a paper that studied this?

I've always trimmed, but doing some recent updated searching, I'm almost seeing an equal proportion of both methods.

Thanks

[Kennels]

anyone have any advice/comments?

[Kennels]

Sorry for the persistence, but I really don't have anyone here to ask for advice on this...
surely someone must have some experience or comments on this? thanks

[shoegame2001]

I have also heard this step suggested, but to me it does not make much sense. Maybe I am misunderstanding, but I don't see how trimming eliminates the priming bias. These are not sequencing errors, but a biased set of true transcripts. Trimming still leaves you with shorter reads from those real, but biased set of transcripts. Anyone with more experience, feel free to correct me.

[schlesin]

In some RNA-seq protocols the error rate of the first few bases is higher, due to mispriming of the random primers. See e.g. Fig. 1 in http://genome.cshlp.org/content/21/9/1543.full

If that is the case, removing those bases might help.

[Kennels]

thanks very much for the replies.

I do agree that removing the biased bases do not really remove the 'biased' reads as these have already been selected for during the sample prep.

I guess my concern is that the reduced complexity in the first 15 bases will affect the assembly process, as many assemblers use distinct kmer sizes from the reads during stages of the assembly. At least that was my understanding.. but then I recently read many papers not doing this trimming, hence my query ...

[Andres_Ribone]

Hi, did anyone see a paper about how trimming (or not) these bases affects assembly? Particularly de novo assembly?

Thanks