Should I trim both R1 and R2 if only R2's quality is poor?

Question

[lakhujanivijay]

This could be a very silly question to ask, but pardon me for that.

I have Illumina paired end data and wanted to trim the reads for quality. Now I have already done fastqc and found that one R2 has poor quality after 225th bases (average read length is 250).

So should I trim both R1 and R2 to ensure paired read lengths are same i.e. 225? or it does not makes a difference if R1 and R2 reads have different lengths? I am going to assemble the reads using de novo approach so how will my decision affect the assembly. Please help.

Let me know if any further information is required from my end.

Answer 1

[igor]

R2 may be worse than R1, but you have bad quality reads in both. If you are going to trim, trim both by quality and not length. Regardless, paired reads should be trimmed together.

Additionally, trim adapters as well, especially if you are using this for assembly.

Answer 2

[piet]

You should trim only those reads which are actually suffering from poor quality. It is quite normal with Illumina, that the second read has (slightly) lower quality than the first read. And hopefully you have prepared your library such that first and second read do not overlap, eg the average insert size is larger than twice the read length.

[GenoMax]

Just wanted to add that the pair of read files (R1/R2) should be trimmed together (using a PE aware trimming program). This would make sure that the files stay in sync in terms of read order (if a read gets removed from one file, its mate is also removed from the other file).