Looking for other CDS changes between the wt and ko samples

1 Comparing my DE-ish data with IGV

The goal of this worksheet is to compare the most changed WT/KO regions of the genome between my differential-expression based examination and what I actually see in IGV. I sort of assume that the differences are going to be due to peculiarities already observed in the WT samples, but lets find out!

1.1 KB664734 first hit

The most different region of all is near the gene ID: ASTE002813

Looking at it in IGV, the reason is immediately obvious. The WT 3 sample has coverage in this region when all other samples (including the other wild types) do not.

first

1.2 KB664942

This is a much more interesting region. It should be noted though that this is a pretty small scaffold, thus the relevant reads may actually be on one of the actual chromosomes. Though I think this is not terribly likely and I am not even sure how to properly test that hypothesis… hmm I guess I can blast a piece of this against the entire genome and see if there is another region which is sufficiently similar. However, the only way that should be relevant is if the mapper (hisat2) mapped the relevant reads differently, which is theoretically possible if we don’t have the random seed set, it randomly assigned the reads to different places on a per-sample basis so that only the KO samples got reassigned differently, and no reads got assigned to this region. So, basically the chances are like 1 in the number of stars – or impossible.

Thus I think this might be a real difference which is related to the difference between the WT and KO samples. Whether this is actually CRISPR I am not sure how to discern.

second

1.3 KB664734 second hit

This looks quite similar to the previous observation. This region at the final 5-10kb of the contig has been lost in the KO samples.

third

1.4 KB664542 first hit

This is much more similar to the first observation. WT sample 1 is very different from all other samples. I would guess that there is 0 chance that this is related to the CRISPR.

third

1.5 KB664488 first hit

This seems an interesting candidate for a funky off-target effect.

fourth

1.6 KB664488 second hit

This seems to be irrelevant. It appears that different samples lost coverage in this region (both WT and KO), but a bit more in the KO samples than WT, so it got picked up. It seems like this entire contig has lower coverage than the others I have looked at, which is presumably contributing to these false positives (there are a bunch more which I am not including).

fifth

1.7 KB664439 first hit

This one is neat, but I would be willing to bet that it is not CRISPR related. Even though all the KO samples are similar in pattern, they share this pattern with WT1.

sixth

1.8 KB665365 first hit

I might stop here, there are 22 more candidates according to my completely arbitrary 4 fold cutoff, but all the ones I looked at are similar in pattern to this one, where one or more WT samples diverged from an otherwise established coverage pattern. Thus I am reasonably confident that there is no significant chance that the differences are CRISPR related.

seventh

I am, however, taking pictures of them so that you may judge for yourself. The naming convention is: \({contig}_\){observation} where config is the name of the assembly’s contig and observation is just the arbitrary order in which I find them.

fourth

pander::pander(sessionInfo())
message(paste0("This is hpgltools commit: ", get_git_commit()))
this_save <- paste0(gsub(pattern="\\.Rmd", replace="", x=rmd_file), "-v", ver, ".rda.xz")
message(paste0("Saving to ", this_save))
tmp <- sm(saveme(filename=this_save))