caseywdunn / comparative_expression_2017 Goto Github PK

At present all the analyses are in manuscript.rmd. This is getting cumbersome to work with as most of the code is stable but increasingly computationally expensive.

Switch to a model where:

• All computationally intensive analyses are run by manuscript_kernel.R. This can be generated by modified knit::purl() output.

• manuscript_kernel.R saves a manuscript.RData file. It has no other output or side effects. This file is not checked into the repo, as it is anticipated to be >250MB in size. It can be moved between computers (eg executed on the cluster and then moved to laptop). The docker code will facilitate consistent environment.

• manuscript_kernel.R can also save intermediate manuscript_checkpoint_XX.RData files to facilitate debugging when the code does not run to completion.

• All code in manuscript_kernel.R will be removed from manuscript.rmd. manuscript.RData will be loaded with load(). The manuscript will contain inline code, plot rendering, and very lightweight analyses.

Cat noted that there are some NAs for species in expression after:

left_join( tip_annotations, by="Ensembl.Gene.ID" )

in https://github.com/caseywdunn/comparative_expression_2017/blob/master/manuscript.rmd

This could be due to errors in parsing tip_annotations or due to missing data in the SEQ fields.