Analyze yeast cytosolic 8x dicodon screen data

[kychen37]

GanttStart: 2022-03-29

Background

Transformed at high efficiency using Bloom lab's liquid recovery protocol: https://github.com/rasilab/rqc_aggregation_aging/issues/98
Library prepped: https://github.com/rasilab/rqc_aggregation_aging/issues/104

Ideas TODO

• Analyze the endogenous fragment data in the same library and see if the codon or amino acid effects can account for their mRNA effects.
• Put the same library in different quality control factor mutants to see their effects on codon and aa mRNA effects.
• Correlate with CSC, AASC, tAI
• Correlate with ribosome pause scores from Weinberg 2016

Analysis Links

• snakemake on cluster: https://github.com/rasilab/lab_analysis_code/tree/master/kchen/analysis/deepseq/20220329_exp51_cyto_8x_dicodon
• analysis in devcontainer: https://github.com/rasilab/rqc_aggregation_aging/tree/deepseq/analysis/deepseq/20220329_exp51_cyto_8x_dicodon

Brief conclusion

• analysis used for publication here and here

[kychen37]

@rasi

• Next week I will go through this data and https://github.com/rasilab/rqc_aggregation_aging/issues/101 more because I want to present it for joint lab meeting

[rasi]

@kychen37 Ok. I remember you showing some data in the short updates, but I don't remember ever us discussing this. Can you update the issue whenever you finish some analysis (I think you used to do this previously?) I can give more useful feedback if I have some time to think about your results than just looking at them for the first time during group meeting.

[kychen37]

@rasi

Code: https://github.com/rasilab/rqc_aggregation_aging/blob/master/analysis/deepseq/20220329_exp51_cyto_8x_dicodon/scripts/plot_alignment_statistics.ipynb

[kychen37]

Prelim analysis, just making a heatmap and starting to look at inserts that are still missing despite good representation:

Code: https://github.com/rasilab/rqc_aggregation_aging/blob/master/analysis/deepseq/20220329_exp51_cyto_8x_dicodon/scripts/plot_insert_mrna_levels.ipynb

[kychen37]

Note to self: my branches are kind of messed up right now, my most recent analyses were done in master (I had switched to doing these analyses in deepseq and forgot, then continued in master). Pulling from master didn't update the deepseq branch, I may need to 'sync' or something, for now the master branch is the most up to date for the scripts of this analysis only

[kychen37]

@rasi Update I will revisit this data next week to see what is pertinent for my committee meeting

[rasi]

@kychen37 Post the figures as Issue comments so that we can come back to them easily (the commit links above are enough to figure out where the figure is).

Why is the data not centered around 0 unlike here: https://github.com/rasilab/lab_analysis_code/blob/master/rasi/analysis/deepseq/20210703_pb_8xdicodon_resequence_grna_mrna/scripts/plot_human_8xdicodon_effects_files/figure-gfm/unnamed-chunk-6-1.png?

[kychen37]

@rasi I've been trying to figure that out but I'm not sure. Could it have to do with the fact that the mean of all the dipeptide lfcs is slightly negative?

[rasi]

@kychen37 The mean and median cannot be this different. First, calculate simple log fold change without bootstrapping and make sure you understand what is going on before doing the bootstrap. The bootstrap is just for error bars.

[kychen37]

@rasi sorry I forgot that my lfc column had already been median normalized when I calculated the median and said it was zero. The median is actually -1.5 and mean is -1.7

[kychen37]

After median-normalization, the median = 0 and mean = -0.19

[rasi]

Either way, the above plot should be approximately centered around 0 if everything is done correctly.

[rasi]

Can you also add horizontal lines for each amino acid similar to what is in Phil's paper?