Cutadapt: Remove Too Short Reads

After removing an adapter sequence from my fastq file the size is reduced from 20 to 13 gb.

When I try to use bowtie with the cut-adapted fastq-file it gives me these two error messages:

Warning: Skipping read (DJDP4KN1:69:C09JPACXX:8:1106:19662:76992 1:N:0:) because it is less than 4 characters long

Warning: Skipping read (DJDP4KN1:69:C09JPACXX:8:1106:19834:76802 1:N:0:) because it had length 0

Not just once or twice, but seemingly a million times.

I suspect cutadapt snipped the reads so that many of the stubs left are too short.

What should I do?

Can cutadapt be configured to not leave reads that are very short?

Should I just ignore the warnings? Is there a way to turn them off?

Does any of this sound fishy, or as if I have done something wrong?

Thanks.

bowtie fastq

• 13,184 views • link

updated 13.7 years ago by Sukhi Singh

• written 13.7 years ago by Click downvote

2 answers

Try running the FastQC toolkit to check the Per base sequence quality graph. See, how the graph looks like. Check a good quality dataset and a poor quality dataset, to see where you data lies.

Also, in cutadapt, you might wanna use the -m paramter. From the manual,

-m LENGTH, --minimum-length=LENGTH Discard trimmed reads that are shorter than LENGTH. Reads that are too short even before adapter removal are also discarded. In colorspace, an initial primer is not counted (default: 0)

So, try keeping at minimum 20-25 depending on which machine was use to sequence it. HiSeq-2000/2500 generally gives better quality if the sequencing itself is not messed up.

• 1 views • link

written 13.7 years ago by Sukhi Singh