andrewjrobinson/rad-loci

A RADseq pipeline based on a read-cluster approach

Rad-loci pipeline

A RADseq pipeline based on a read-cluster approach

Installation

Dependencies

1. VSearch (https://github.com/torognes/vsearch) [Version 1.1.3 was used when testing]
2. Parallel (http://www.gnu.org/software/parallel/) [Version 20140722 was used when testing]
3. Python (https://www.python.org/) [Version 2.7.6 was used when testing]
4. Numerous shell tools (awk, basename, bc, bzcat*, cat, cd, grep, ln, mkdir, paste, pwd, rm sed, sort, tail, uniq, wc, zcat*) *optional commands

NOTE: there is a command in the bin directory that checks that it can find all necessory commands called 'check-dependencies'

Rad-loci

1. Download the latest release
2. Extract archive
3. Add bin directory to your path

e.g.

# Download the source (and unpack)
wget https://github.com/molecularbiodiversity/rad-loci/archive/v0.4.tar.gz
tar xf v0.4.tar.gz

# Optional: you may want to move the source somewhere else.  If you do you
# will need to cd to the directory containing it too.
#mv rad-loci-0.4 /usr/local
#cd /usr/local

# Add to path (you may want to put this in your ~/.bashrc file to you don't
# need to do it each time).
# use 'pwd' command to work out your working directory (as you will need to
# hard-code it if you use the ~/.bashrc option)
export PATH=$PATH:$PWD/rad-loci-0.4/bin

# Check dependencies were ok
check-dependencies
# this should output "Success: everything found in your path" at end

Quick start

1. Make a directory called 'samples' in your current directory and copy/symlink all your sample fastq or fasta files into it
2. Run 'rad-loci-settings settings.conf' program to create a new settings file in your current working directory. Optionally change any of the settings as required
3. Run 'rad-loci settings.conf' command. It will output a lot of progress information on the terminal. It will take a while to process the data based on how much input data is provided so it is best if you run in with nohup or as an HPC job.

e.g.

# create a directory for this experiment
mkdir experiment
cd experiment

# create a directory for the samples (input)
mkdir samples
cd samples
# copy or symlink the sample fastq/fasta files
#ln -s /some/path/to/sample/files*.fq .
#cp /some/path/to/sample/files*.fq .
cd ..

# create settings
rad-loci-settings settings.conf
# edit the settings as necessory
#nano settings.conf
#vim settings.conf

# run the pipeline (with nohup as it might take a while and
# we don't want it to terminate if our connection drops)
nohup rad-loci settings.conf &

# watch the output
tail -f nohup.out
#Ctrl+C to exit this view as it will go for ever

Processing steps

1. Create the catalog of potential loci and prepare samples for analysis
2. Merge (cluster) similar sequences from catalog
3. Filter clusters to those that have 2 to 16 different members (<2 = non-informative, >16 = multi-mapping)
4. Align sequences to filtered catalog (to find the which alleles map to each)
5. Refilter catalog to those that contain 2 to 16 alleles
6. Make database of all allele sequences (FastA)
7. Map sequences from each sample (100% identity) to allele database and count copies
8. Merge counts from each sample into single TSV file
9. Filter Loci based on:
10. Max 2 alleles per sample per loci
11. Min 2 significant alleles and
12. Min proportion of samples with data for loci
13. Call genotype for each sample and output as a structure (.stru) file

optional

1. Make random selection of loci
2. Call genotype into migrate format