The Bismark Boat

I have been working through Bismark with a few Crassosstrea virginica datasets. This includes the BS data from the 2015 Oil exposure experiment, OA exposure - gonad tissue (OAKL), and a full suite of library preps via Qiagen.

A few take aways is are 1) use the -u feature to work out analysis on subset, 2) the --score_min variable is important for allowing for mismatches, 3) the working directory approach seems to work good given the number of files.

I have created a few notebooks for running most of the Bismark pipeline.

Most cells do not need much attention besides this one..

find /Users/sr320/Desktop/trim14/zr2096_*R1* \
| xargs basename -s _s1_R1_val_1.fq.gz | xargs -I{} /Applications/bioinfo/Bismark_v0.19.0/bismark \
--path_to_bowtie /Applications/bioinfo/bowtie2- \
--genome /Users/sr320/Dropbox/wd/18-03-15/genome \
--score_min L,0,-1.2 \
-u 10000 \
-p 2 \
--non_directional \
-1 /Users/sr320/Desktop/trim14/{}_s1_R1_val_1.fq.gz \
-2 /Users/sr320/Desktop/trim14/{}_s1_R2_val_2.fq.gz \
2> bismark.err

Here you need to be aware of file naming structure for basename and of course where the files are.

As far as some results. For the OAKL samples sample 2 seems a bit off.


Alignments seem good. Here is one of the files.


The M-bias plots bother me a bit.


Plotting bedgraphs does not immediatly show OA impacts


But this is just mapping 10k reads, going up to 100k is evidient.


The Qiagen library analysis is indicative of the different prep methods



This might be more telling..


another way of looking at it.


though is worth pointing out these library have not been trimmed and likely need to be done in specific ways.

Written on April 15, 2018