Germline_Benchmark

Brief introduction

The inputs to hap.py are two VCF files (a "truth" and a "query" file), and an optional "confident call region" bed file.

NOTE: bed files with track information are not supported, all input bed or bed.gz files must only contain bed records.

You can find more detailed information Hap.py User's Manual

Processing Steps

Running benchmark

run_benchmark_hg38_parallel_computing.sh

Bed intersect and selectvariant for WES

bedtool intrersect allows one to screen for "overlap" betwenn two sets of genomic features. Select a subset of variants from VCF file using GATK tool.

run_bedintersect_select.sh

ROC curves

Hap.py can create data for ROC-style curves. Normally, it is preferable to calculate such curves based on the input variant representations, and not to perform any variant splitting or preprocessing.

run_rocplot.Rscript.sh

Any set of hap.py runs can be plotted like this using the script run_rocplot.Rscript.sh

WGS benchmark Result

SNVs

INDELs

WES benchmark Result

SNVs

INDELs