Question about RRBS mapping efficiency

Question

[hxlei613]

I've been working on RRBS data these days. Raw data were processed using Trim_galore tool, in order to remove adapter and low-quality reads. The clean data were mapped to the reference genome using the Bismark alignment tool with default parameters(--bowtie2 --directional). Final Alignment report is like this:

Sequence pairs analysed in total:       17109021
Number of paired-end alignments with a unique best hit: 10941462
Mapping efficiency:     64.0%
Sequence pairs with no alignments under any condition:  3219358
Sequence pairs did not map uniquely:    2948201
Sequence pairs which were discarded because genomic sequence could not be extracted:    0

Total number of C's analysed:   539400501

Total methylated C's in CpG context:    28032910
Total methylated C's in CHG context:    2657563
Total methylated C's in CHH context:    5817321
Total methylated C's in Unknown context:        0

Total unmethylated C's in CpG context:  90723174
Total unmethylated C's in CHG context:  147210060
Total unmethylated C's in CHH context:  264959473
Total unmethylated C's in Unknown context:      1

C methylated in CpG context:    23.6%
C methylated in CHG context:    1.8%
C methylated in CHH context:    2.1%
C methylated in unknown context (CN or CHN):    0.0%

Is the resualt ok ? I want to know wether the 64% mapping efficiency is ok for RRBS data and why. And why is it so low compared to other NGS ?

Thank you very much.

Answer 1

[Devon Ryan]

The methylation percentages seem pretty reasonable (normally you expect CHG and CHH methylation to be 1 or 2 percent). Not knowing what species this is, it's hard to judge if 64% alignment is good or now. For mouse I normally got higher (80-90%), but either way it's not terrible and with RRBS you probably have enough coverage as is.

BTW, RRBS has a lower alignment rate than other things because you only have 3 nucleotides to work with when doing the alignment. Consequently, there are a lot of reads that map to a million places and are consequently just tossed.

[hxlei613]

I forgot to say ... it's human.

[Devon Ryan]

That seems a bit of a low alignment rate to me then. What happens if you instead use bwa-meth? Just try that on one sample and see what sort of alignment percentage you get out.

[hxlei613]

Ok,I will try. By the way, what is the common RRBS alignment rate for human ?

[Devon Ryan]

I would expect 80-90%.

[mrodrigues.fernanda]

Devon, just a follow up on this thread. Is a mapping efficiency between 35% and 45% too low for bovine?