Compare across samples to improve predictions of low-frequency variants

[jeffreybarrick]

• Create files that dump information about alignments into a format that can be easily read in for testing different algorithms. Use Ara-1 mixed population datasets as a test case.
• Create an algorithm for calculating the confidence that the fraction of a base variant at a position is different in one mixed population sample versus a set of samples (including clonal samples).
• Calculating credible/confidence interval for the fraction of a variant in a population.

[jeffreybarrick]

Use example data from:

Deatherage, D. E., Traverse, C. C., Wolf, L. N., Barrick, J. E. (2015) Detecting rare structural variation in evolving microbial populations from new sequence junctions using breseq. Front. Genet. 5: 468.

It consists of 24 mixed population samples described here:
https://github.com/barricklab/LTEE-Ecoli/tree/master/LTEE-mixed

For a smaller dataset to analyze, use the fabR gene region (REL606:4140000-4142000)

samtools mpileup --max-depth 1000000 -r REL606:4140000-4142000 -o fabR.pileup.tsv -f 03_Output/Ara+5_500gen_REL772/data/reference.fasta `ls 03_Output/*/data/reference.bam` 

Output available for download here: https://barricklab.org/release/breseq_development/issue221/fabR.pileup.tsv.gz

They are in pileup format: http://samtools.sourceforge.net/pileup.shtml

[jeffreybarrick]

@dgauraang The new pileup that compares just two samples is available for download here:

https://barricklab.org/release/breseq_development/issue221/two-sample-pileup.tsv.gz