Thoughts on a few bad tiles per sequencing run

Question

[Biogeek]

I was doing some reading and came across the Macmanes article discussing how minimal trimming is better than stringent trimming for mapability in RNA-Seq reads.

For Trimmomatic I used such settings:

'adapters/TruSeq3-PE-2.fa:2:30:10 HEADCROP:13 LEADING:5 TRAILING:5 SLIDINGWINDOW:4:5 MINLEN:25'

Whilst my sequence per bp quality is high, and my per base sequence content also. My per tile quality was flagged red due to 2 bad tiles (image link below). What would one do, I've seen flow cells with major tile problems across the whole cell. Do 2 cells make a big impact on the data, is there anyway to remove these, or would one bother given the other test parameters passed?

Additionally my kmer content is flagged;however I've read this is typical and not a problem?

Thanks.

https://postimg.org/image/fbtwek7wl/

[GenoMax]

Just follow your standard protocol. A couple of bad tiles are not going to cause major issues.

Answer 1

[Devon Ryan]

Don't worry about a couple bade tiles. The fastQC kmer thing always fails in RNAseq, ignore it.