The goal of the analysis is to find regions the genomes of Common and Thrush nightingales, which could contribute to the recent speciation event. We used SNP variants, per site Fst and Dxy as indicators of possible speciation, alignment to Zebra finch as a proxy for genomic location of the variants, floating windows to scan along chromosomes, and bootstrap to detect significanly higher values.
tidyversedata.tablefor fast interval operationsgtoolsfor mixed sort of chromosome names
To get more reliable results, variants should be filtered on some biological significance. Filter out nightingale contigs mapped to too wide range in the zebra finch, filter out variants on some quality threshold, for Dxy filter variants with too few called individuals.
Cover all chromosomes with equally spaced intervals. Use at least 1000 intervals
for the longest chromosome (?). Most of the calculations is done in pure data.frames,
with the help of dplyr, tidyr and visualised with ggplot2.
We're using data.table::foverlaps package to find the variants belonging to each window.
Bootstrap was performed via permuting the values (Fst, Dxy) assigned to each genetic variant within chromosomes, the variants remain at original positions. We sampled 25,000 permutations, taking the maximum of the windowed measure at given grid point as the threshold value for 'maximal Fst/Dxy' arising by pure chance in the given variants and chromosomal arrangement.
Areas where both Fst and Dxy exceeded the threshold found in the bootstrap
were taken as putative 'islands of speciation'. We used biomaRt to get
information on genes found in those regions.
After trying several online tools and having non-reproducibility issues, we decided to
run the analysis on simple KEGG pathways on our own. Using simple gene lists in data.frames
and Fisher's exact and Binomial tests.