nf-core radseq design

[remiolsen]

This is built on https://github.com/remiolsen/NGI-RADseqQC

This is a major rewrite to make this pipeline harmonious with nf-core, update the tools used (Stacks 2.0, remove read-joining, etc). Also a few other points that have been on my wishlist for improving usability.

Main tasks

• Think of a cool new name
• Make cookiecutter template
• Port over some of the processes from https://github.com/remiolsen/NGI-RADseqQC
• Use Stacks 2.0
• Write a tool to scrape the Stacks logfiles for useful stats
• Others

Core pipeline tasks

• Remove FLASH
• Make a dockerfile. Is Stacks 2.0 on bioconda?
• Write a python script to parse denovo stacks to get: coverage, raw # sample loci, catalog loci per sample, "shared" loci histogram. Parse process_radtags also?
• Make a MultiQC configuration to import this data
• Get publically available data from ENA. Make proper test data.
• Make a MultiQC module for Stacks >= 2.0

Polish

• Make a GH release
• Documentation, documentation, documentation
• Travis-CI
• Python3 support for in silico digest helper script

Others -- Stretch goals

• Think about what output files stacks should be creating by default.
• Let the user specify which output files to create -- Nah the defaults are probably fine -- Nuh-uh we need more!
  • genepop
  • structure
• Scripts for running the Stacks web UI -- It's been removed in v >= 2.0
• Pick a set of “best practice” parameters for Stacks and run all of these.
• Clearly report r80 statistic of each run, i.e # of polymorhic loci shared by at least 80% of individuals in the population -- http://doi.org/10.1111/2041-210X.12775
• Support running Stacks with a reference genome
• Support for premade population map file
• Support for already processed reads (skipping trimming and process_reads)
• Option to not output trimmed and/or processed fastq files