Polar Star

This code analyzes the read depth along long-read contigs and finds outliers.

Dependencies

The Polar Star pipeline installs several common bioinformatics tools:

1. Htslib
2. Samtools
3. Bedtools
4. Vcflib
5. Minimap2

These codebases have serval system library dependencies. There is a shell script that installs the dependencies (for Amazon EC2 instance). If you’re not using Amazon EC2 the setup script should help you figure out which libraries you need.

Polar Star installs specific commits of code so that the pipeline won’t be broken as command line tools change.

The pipeline is run by snakemake: http://snakemake.readthedocs.io/en/stable/

Setup

edit the config.json file.

{
	    "threads"     : "4",             #the number of threads available 
	    "opts"        : "-x map-pb -a",  #minimap2 options (advanced)
	    "name"        : "sample_name",   #output prefix
	    "lib"         : "la",            #a library name for the bam (most can ignore)
	    "low_depth"   : "2",             #low depth cutoff
	    "times_mean"  : "3",             #high depth cutoff = mean * (times_mean)
	    "fasta_name"  : "PGA1_jelly_arrow_pilon_pallidicuale.gapsplit.fasta", #full fasta name
	    "fastq_names" : "m54120_170710_225913.subreads.fasta,m54120_170711_085749.subreads.fasta,m54120_170711_190715.subreads.fasta" #comma separated list of long-reads to align
	}

Running

 snakemake -p -s Snakefile

What is Polar Star doing?

1. Aligning the long-reads to the assembly.
2. Caculating read depth at every base.
3. Smoothing the read depht in a 100bp sliding window.
4. Merging regions of high (_hd), low (_ld), and normal read depth (_nd).
5. Creating a new fasta that is broken.
6. Generates simple statistics.

FAQ

Q: What if i don't want to break on low read depth?

• A: set "low_depth" value to -1.

Q: How is the high depth caculated?

• A: Polar Star calculates the mean, then multiplies the value in "times_mean".

Q: How did you come up with the "times_mean" default?

• A: Read depth can be modeled with a negative binomial distribution, but I was too lazy. I'm using a poisson. So the mean = the standard deviation.

Q: How can I reject breaks.

• A: Edit the bed file that is the input for the "getFasta" rule.

Q: How can I add breaks.

• A: Edit the bed file that is the input for the "getFasta" rule.

Q: I already have an alignment, how can i use that?

• A: Snakemake is smart. Check the input to the "meanDepth" rule and match the file name. You'll need to change the config.json so the names match.