Best practice for de novo assembly

Question

[codsorre]

Hello all, I am new to de novo assembly and working with sequencing data in general. I am working with some paired end sequences. I just ran them through trimmomatic to remove adapters and am left with the forward and reverse paired sequences, and the forward and reverse unpaired sequences. When running trinity for de novo assembly, would it be the best practice to just assembly using the paired sequences for all samples I am analyzing, or should I include the unpaired sequences as well? What would be the pros or cons of each?

Thanks

Answer 1

[Michael]

Trinity has a switch to run trimmomatic inside its pipeline. However, if you ran it already, I don't see much of a problem in using the unpaired sequences as well. Do you see many unpaired fragments, I normally see only a tiny fraction. If there are too many that might indicate a problem with the library. Stil, unpaired fragments will provide coverage and kmer information for building the De Bruijn graph. If in a later step paired reads are required for orienting contigs, the unpaired will be disregarded.

[codsorre]

That makes sense, thanks. Relatively speaking, how many unpaired fragments is too many? Roughly about 5% seem to be going to unpaired