Pipeline for Transposon Expression Reanalysis (PTERA)

Dependencies

Using mamba 0.9.2 (or conda 4.8.4) install the following:

mamba create -n ptera_v2 \
	python=3.9.6 snakemake=6.7.0 peppy=0.31.1 sra-tools=2.11.0 tabix=1.11 \
	r-tidyverse=1.3.1 r-tidymodels=0.1.3 r-sqldf=0.4_11 \
	xsv=0.13.0 bioconductor-deseq2=1.34.0 bioconductor-preprocesscore=1.54.0 \
	bioconductor-edaseq=2.26.0 samtools=1.14 bioconductor-eisar-1.6.0 bioconductor-bsgenome=1.62.0 \
	r-wgcna=1.69

Make sure you run pre-install the macs env 1 time via basilisk.

basilisk::basiliskStart(MACSr:::env_macs)

Singularity must also be installed and accessible in this environment. Depending on your system this can be done via conda. For HPCs admin privileges may be required.

Other dependencies are handled at runtime by singularity.

Where the data comes from

Accessions are included in the pepfiles for each subworkflow. The primary data (DGRP RNA-seq) was generated by downloading all metadata for PRJNA483441 and processing the metadata file with a custom R script (see workflow/scripts).

Other data were similarly pulled from SRA/GEO.

Prerequisite data resources

This pipeline requires a resources folder which should be downloaded and placed in the pipeline directory (level with config/ for example).

Setting up new subworkflows

New data can be processed in subworkflows. Each sample table should resemble the SRA run selector export. Recommended to clean this table appropriately with a script that can be saved in the workflow/scripts directory of each subworkflow.

Each table should have at a minimum the following columns:

sample_name
LibraryLayout
Experiment
Run
Library

ChIP-seq data should additionally have an input column.

Test Mode

Note that by default RUN_TYPE=TEST for the resource intensive workflows. This is useful for testing individual pipeline components (i.e. subworkflows or the initial portions of the main workflow), but due to the difficulty of making sure equivalent strain samples are always processed in the test dataset for each subworkflow, test mode is not guaranteed to complete the full workflow (main wf + subworkflows). Instead use RUN_TYPE=FULL.

Running the pipeline

snakemake --profile <your profile> --use-conda --use-singularity --use-conda -j 999 -kp \
	--config RUN_TYPE=FULL -n

If disk space is at a premium, consider running with --prioritize salmon_quant_se_vanilla to enforce creation of the quant output and deletion of temp files for each sample as fast as possible.