Do I have to trim anything before I run MAGeCK?

Question

[leranwangcs]

Hi,

I fit my raw data (adapters have been removed) into MAGeCK but always get low mapping rates (around 35% - 40%). I wondered if there is any prior steps like trimming that I skipped so that caused this low mapping rates?

Thanks! Leran

[ATpoint]

Depends. I personally prefer to trim the reads to exactly the barcode sequence and then align this with bowtie2 manually against the barcode fasta file. What kind of library/plasmid did you use and did you check https://sourceforge.net/p/mageck/wiki/Home/#optional-step-3-determine-the-trimming-length-and-sgrna-length ?

I run bowtie2with:

--end-to-end --very-sensitive --rdg 10000,10000 --rfg 10000,10000 --mp 10000,10000

and then use featureCounts to get the count matrix, requiring a minimal MAPQ of 42 which is the highest MAPQ bowtie2 assignes to a read. Together with the strict parameters from above this then eventually counts only reads that exactly align without any mismatch.

[leranwangcs]

Thanks for reply!

Our library is pXPR_502. Also by "barcode sequence" do you mean the guide RNA sequences?

Thanks! Leran

[ATpoint]

In most CRISPR/shRNA applications you sequence a barcode that is attached to next to the sgRNA/hairpin sequence to avoid sequencing regions that tend to form secondary structures. Something like this here downstream of the sgRNA.

I do not know your system in detail. MAGeCK is pretty robust and mature, but it is always good to double check results with an independent method, here by aligning the trimmed reads against the barcode library.