Merging reads for denovo assembly

[Niranjanks]

Hi there

I have MiSeq data and I wish to perform plant De novo assembly.

I removed the PCR duplicates using FastUniq and now I wish to form a metagenome merging the overlapping reads into longer ones. Afterwards I shall do a denovo assembly using Discovar.

I read this blog here: http://thegenomefactory.blogspot.in/...aired-end.html
Although it isn't updated I have also read about other tools like BBMerge.

My question is: I read a lot of conflicting opinions on whether this merging of overlapping reads should even be done. Are there any solid benchmarks on which available software is the best for this role?

I want an assembly that preferably does not tolerate false positives.

[westerman]

The case for overlapping depends on the assembler you use. Which assembler to use not only depends on your species (what works well with a bacteria may not work well with a plant) but also on your computation resource (mainly memory) and the type of reads. For example Discovar required a single 250bp paired end library. Any other library will not work.

Since you have already decided to use Discovar then do not do merging. It will not work with merged reads.