Fixed bugs related to library normalization with no_dedup, replicate formatting, and DESeq

[epehrsson]

Changes

Fixes bugs related to library normalization with no_dedup option, replicate formatting, DESeq (single-sample group check and log2FC parameter). Also allows local library installation as workaround for installation issues in shared directory.

Issues

• fixes deseq2 sample check is failingi #126
• fixes Error in rule bam2bg when running "no_dedup" option #123
• fixes Dryrun fails due to replicate formatting in samples.tsv #119

PR Checklist

(Strikethrough any points that are not applicable.)

• This comment contains a description of changes with justifications, with any relevant issues linked.
• Update docs if there are any API changes.
• Update CHANGELOG.md with a short description of any user-facing changes and reference the PR number. Guidelines: https://keepachangelog.com/en/1.1.0/

[kelly-sovacool]

@epehrsson can you update the changelog (CHANGELOG.md) under the development version header with a description of these changes?

You can use something like this (or copy-paste this directly)

## CARLISLE development version

- Bug fixes (#127, @epehrsson)
    - Removed single-sample group check.
    - Increase memory for DESeq.
    - Ensure control replicate number is an integer.
    - Fix FDR cutoff misassigned to log2FC cutoff.
    - Improve `no_dedup_nreads_genome` variable name.

## CARLISLE v2.5.0

[kelly-sovacool]

We should update the docs here

CARLISLE/docs/user-guide/preparing-files.md

Lines 153 to 160 in 31ecafd

	### 2.2.2 Contrast Manifest (OPTIONAL)
	This manifest will include sample information to performed differential comparisons.
	An example contrast file:
	| condition1 | condition2 |
	| --- | --- |
	| MOC1_siSmyd3_2m_25_HCHO | MOC1_siNC_2m_25_HCHO |

With something like this:

**Note**: you must have more than one sample per condition in order to perform differential analysis.
Please ensure each condition contains more than one sample.

[kelly-sovacool]

Thanks for contributing this, @epehrsson! See my comments above for suggested changes

• changelog update
• docs update
• remove commented-out code

I think @kopardev should also take a look since he has worked on this pipeline more than I have, as I may have missed something.

[kelly-sovacool]

@kopardev Maybe now is the time to go ahead and create a container for the DESeq step so we can get rid of the code related to installing R packages? (#128)

[epehrsson]

To add more context, the commented-out code allows the user to use their own R library directory instead of the central pipeline directory. Without this edit, rule DESeq fails due to missing packages (within _diff_markdown.Rmd, setup chunk, the CARLISLE_HANDLE_PACKAGES call). In particular, it seems unable to load some ChIPSeeker dependencies. It also no longer handles ELBOW installation. However, the user must install some libraries manually with this edit.

[kelly-sovacool]

@epehrsson I chatted with Vishal and he agrees we should switch to using containers. For now, I think the best thing to do would be undo the changes related to the R package installation / lib path code. In a separate PR I'll get rid of that code and add a container.