Yaamini’s Notebook: October 2018 Goals


September Goals Recap:


  • I finished addressing reviewer comments! The manuscript is almost finalized for resubmission.


  • I presented my preliminary results at PCSGA!
  • Worked on in compGO and DAVID, but my methods are most likely wrong
  • Didn’t make much headway writing because the gene enrichment stuff was confusing enough

Gigas Broodstock:

October Goals


  • Re-submit. That. Paper.


  • Address minor reviewer comments
  • Address major comments
  • Try using multivariate methods to explore relationship between environmental data and protein expression data
  • Upload proteomics data to online databases
  • Reformat the paper repository

Gigas Broodstock:

  • Finish extracting broodstock DNA
  • Identify sequencing protocol
  • Prepare samples for bisulfite sequencing


  • Rework bismark and methylKit to solidfy analysis parameters
  • Present more solid findings to Eastern oyster methylation project group
  • Update methods and results
  • Fix gene enrichment methods

// Please enable JavaScript to view the comments powered by Disqus.

from the responsible grad student https://ift.tt/2xWEVis

Running Bismark

~/tools/bismark_v0.17.0/bismark relative/path/to/prepared/genome –bowtie2 –score_min L,0,-0.6 -1 blah_R1.trim.gz -2 blah_R2.trim.gz -p 12 –bam

Of all the commands here, the –score_min stands out as the unexplainable non-default. This exact command was suggested by the program creator in one of the forum posts (http://seqanswers.com/forums/showthread.php?t=40496) when someone complained they obtained very low mapping rates for their data.

“you haven’t mentioned the read length or the genome you are working with, is it a good assembly or still quite rough? Do you expect many SNPs? Do you expect insertions or deletions in the reads? In any case I would try running Bismark in Bowtie 2 mode (–bowtie2), which allows mapping with InDels and also has a function (–score_min) for the number of mismatches and/or InDels that scales with the read length. This is set very stringently by default (L,0,-0.2), but could be relaxed somewhat to see if this helps your mapping efficiency. As a guideline, –score_min L,0,-0.2 would allow ~3 mismatches for a 100bp read, –score_min L,0,-0.6 would allow up to 10 mismatches and/or Indels. If you just want to test these things you can use the -u option to just use a subset of the reads.”

–score-min controls how “loose” the mapping should be–and from testing, I can vouch that this cutoff strikes a good balance between sensitivity and accuracy. –score_min is a Bowtie2 parameter, read its docs if you want to know what it means EXACTLY. To pre-empt a bit of confusion if you actually DO read the docs, Bowtie2’s default –score-min is L,-0.6,-0.6 for end-to-end mapping. This default setting is OVERRIDDEN by bismark–bismark sets this parameter as L,0,-0.2 if one does not override its defaults. We’re basically overriding bismark’s defaults to bring it more in line with bowtie2’s default setting.

In this pipeline, –score_min is the BIGGEST determinant of the results. This is the one thing I’ll optimise if my results look weird. If mapping rates are really low (< 20%), I suggest using L,0,-0.9 or L,0,-1.2.