ryhui / imputation-pipeline

imputation-pipeline

This repository contains scripts used in an imputation pipeline for low-coverage (0.05-1x) ancient genomes.

imputation_pipeline.snakefile

Default pipeline, including down-sampling, genotype likelihood calling in ATLAS, and imputation in Beagle. It is written as a snakemake workflow, but bash commands can be easily extracted from the "shell" sections.

local_imputation_pipeline.sh

(Author: Eugenia D'Atanasio)

Pipeline used for local imputation when only genotypes at certain positions are of interest.

pmd_filter.py

Filters out GLs that could result from deamination (C -> T changes).

compare_accuracy.py

Compares imputed and true genotypes to produce the accuracy table for various allele frequency bins.

imputation-pipeline's Issues

Quick Question

How long do you approximate it would take your pipeline to run on a single human genom ~1x coverage?m
thank you-
je,

ATLAS arguments

Should the "sites=" argument to ATLAS be "alleles=" instead?

I'm running with ATLAS 0.9 and when try to specify the known sites with "sites=" as in the pipeline, it seems to output all sites that are covered by at least one read - but when I run with "alleles=", it outputs just the specified sites as expected. The ATLAS documentation describes the "alleles" argument but not "sites": https://bitbucket.org/wegmannlab/atlas/wiki/Variant%20Discovery%20Tools:%20call.