WGBS Analysis

Testing bismark parameters for WGBS data

We’re 18 days out from PCSGA, which means I should start analyzing my data! In February, I pooled C. gigas samples from ambient and low pH treatments for Whole Genome Bisulfite Sequencing (WGBS). I’m planning on processing the data to look at differential methylation between treatments so I can justify future sequencing.

Determining parameters

The first step in this process is aligning samples to a bisulfite genome with bismark. I started this Jupyter notebook to test different alignment stringencies with a subset of the data before running the full bismark pipeline. Sam already trimmed my files and prepared the bisulfite genome, so that’s two things I didn’t have to do! I downloaded the bisulfite genome from this link and trimmed samples from this link (names start with YRV). I would have run bismark without downloading first, but owl is super slow I’d rather just have them on genefish.

Once I had the files dowloaded, I referred to my C. virginica script for parameter testing. The last thing I wanted to do was confirm my files were paired and non-directional. I was pretty sure they were paired because there were two files for each sample. I wasn’t sure if they were non-directional. I posted this issue to confirm. After running different alignment stringencies, I found that score_min = L,0,-0.9 gave me about 61% alignment. Less stringent settings gave me 68% alignment, while more stringent settings gave me 47% alignment. I decided to stick with 61% alignment because it was the biggest jump between settings, and I didn’t gain much from less stringent settings.

Running bismark on Mox

Now that I had my alignment settings in order, I could run a full alignment on mox. I modified a previous script for bismark on Mox with my C. virginica samples to use with my C. gigas samples. I saved the revised script here. I used the following commands to move files over to Mox:

rsync --archive --progress --verbose yaamini@*fastq.gz .#Transfer analysis files
rsync --archive --progress --verbose yaamini@ . #Transfer bisulfite genome to Mox
rsync --archive --progress --verbose yaamini@ . #Transfer script to USER directory, not scrubbed directory

To run the script:

sbatch 2019-09-03-Bismark.sh

It’s been a while since I last used Mox, so I encountered this error. Turns out --workdir is no longer used! I changed it to --chdir. I learned a new verison of bismark was installed, so I changed version 19 to 21 in the script. Once I made those changes, I reran the script! The job number was 1266351.

Written on August 30, 2019