Chip-seq reads' trimming

Question

[marina-orlova]

Hi all

I have chip-seq reads of bad quality (dm3) - 29,772,199 reads:

Tried to trim it with fastq_quality_filter (command line fastq_quality_filter -q 20 -p 100 -i infile.fastq -o outfile.fastq) and got only 410,775 reads:

Is that OK? Otherwise, what should I do to these data? I suppose I shouldn't just push it to bowtie for alignment..

Answer 1

[Chirag Nepal]

From fig it looks the quality of your sequence is not good. Currently u are throwing out most of your data by trimming. 410775 is too little reads to push to bowtie.

First check if you have primers/adapters attached to beginning/end of sequence. If so, remove them. Look at over-represented sequence, take few of those, blat them to genome individually and see what portion of the sequence maps and what doesn't. That should give you idea about primer/adapter sequence.

Or trim the last 5 or 10 nucleotide, and analyze how mapping differs ? compare with different parameters of trimming. You need to play a bit with your data.

Answer 2

[Brian Bushnell]

Bowtie is not very tolerant of low-quality reads. If you want to use this data, I'd suggest aligning with BBMap, and using the raw untrimmed reads.