Using Pilon after MaSuRCA Assembly

[jpummil]

So, lots of good threads on using Pilon...but my question is specifically about the use of Pilon after using MaSuRCA for assembly. MaSuRCA specifically states that the unmodified, raw reads be used without any sort of manipulation:

"IMPORTANT! Do not use third party tools to pre-process the Illumina data before providing it to MaSuRCA, unless you are absolutely sure you know exactly what the preprocessing tool does. Do not do any trimming, cleaning or error correction. This will likely deteriorate the assembly."

Again, no problem for the assembly portion. BUT....when I move to the Pilon polishing step(s), do I want to try and use the raw reads as they are for polishing?!? Seems that, depending on how Pilon interprets that data, it might introduce some of the low quality aspects back into the assembly? Or does the mapping of the raw to the assembly to create the .bam resolve any such possibilities?

[Gopo]

Personally, I would do quality and adapter trimming (and maybe error-correction) of the raw Illumina reads as MaSuRCA pretty much does this from what I understand and use the quality and adapter trimmed reads for polishing with Pilon.

[jpummil]

Kind of leaning that way for sure. I dug around a bit in the MaSuRCA documentation and have identified and located the "corrected" PE data created during assembly:

pe.cor.fa - error corrected PE reads. The ordering of the reads is arbitrary, but the pairs are guaranteed to appear together. No quality scores

So, the file looks more like merged PE data and is now a .fasta, though I don't know if Pilon was using the quality scores or not...

I figure it's worth a try anyway....maybe compare it with the raw PE data and see what looks better in a Quast analysis...

[jpummil]

Haha! Well, THAT didn't work! BWA did OK with the indexing, etc...but Pilon quickly bailed:

Suppressed: java.lang.IllegalStateException: Inappropriate call if not paired read

I'll go back and manually quality trim and error correct the raw files.