selfdecode / rd-imputation-accuracy

Phasing and genotype Imputation comparison. Have been evaluated: BEAGLE 5.4, EAGLE 2.4.1, SHAPEIT 4, MINIMAC 4, IMPUTE 5, using accuracy metrics like: IQS(Imputation Quality score), r2 (Pearson correlation), Concordance.

A comparative analysis of current Phasing and Imputation software

Source Code

Usage:

The help shows all the required arguments listed above, plus optional arguments.

Usage: ./Imputation_score.sh -i <input.vcf.gz> -r <ref_file> -t <4> -o <output_name> -c <20>
Use -h or --help to display help.

author: SELFDECODE
contact email : [email protected]

SelfDecode pipeline to analyze multiple phasing and imputation softwares simultaneously 

Parameters:

      -h|--help
              show help
      -i|--input
              input file
      -r|--reference
              full path reference file without extension.
      -t|--threads
              number of cpus to use.
      -o|--output
              output prefix. No extension.
      -c|--chr
              chomosome to analyze allowed [1-22 X]. NO chr prefix.
      -ibeagle|--imp_beagle
              skip Beagle imputation
      -pbeagle|--phase_beagle
              skip Beagle haplotype estimation
      -impute5|--impute5
              skip Impute imputation
      -shapeit|--shapeit
              skip ShapeIT Phasing
      --minimac|--minimac
              skip Minimac imputation
      -eagle|--eagle
              skip Eagle phasing
      -bigref|--BIGREF
              use this option if you get memory allocate error during accuracy evaluation

[base] Usage: ./Imputation_score.sh -i <input.vcf.gz> -r <ref_file> -t <4> -o <output_name> -c <20>
[skip] Usage: ./Imputation_score.sh -i <input.vcf.gz> -r <ref_file> -t <4> -o <output_name> -c <20> -ibeagle no -impute5 no
[memo] Usage: ./Imputation_score.sh -i <input.vcf.gz> -r <ref_file> -t <4> -o <output_name> -c <20> -bigref yes

Before running the example - change paths for software

#SOFTWARE PATH (EDITABLE PART)
time="/usr/bin/time -f"
repo="/home/ec2-user/adriano/imputation/phase2/software"
eagle2="${repo}/eagle2.4.1/Eagle_v2.4.1/eagle"
shapeit4="${repo}/shapeit4/shapeit4-4.2.1/bin/shapeit4.2"
beagle5="java -Xmx8g -jar ${repo}/beagle5.2/beagle.29May21.d6d.jar"
bref3="java -Xmx8g -jar ${repo}/beagle5.2/bref3.29May21.d6d.jar"
imp5Converter="${repo}/impute5/impute5_v1.1.5/imp5Converter_1.1.5_static"
miniConverter="${repo}/Minimac3Executable/bin/Minimac3"
impute5="${repo}/impute5/impute5_v1.1.5/impute5_1.1.5_static"
minimac4="${repo}/minimac4/Minimac4/build/minimac4"
simpy="/home/ec2-user/adriano/git/rd-imputation-accuracy/bin/Simpy.py"
imputation_accuracy="/home/ec2-user/adriano/git/rd-imputation-accuracy/bin/imputation_accuracy.sh"
#GENETIC RECOMBINATIO MAP PATH
map_eagle2="${repo}/eagle2.4.1/Eagle_v2.4.1/tables/genetic_map_hg38_withX.txt.gz"
map_beagle5="/home/ec2-user/adriano/imputation/phase2/genetic_map/plink.chr${CHROMOSOME}.chr.GRCh38.map"
map_shapeit4="/home/ec2-user/adriano/imputation/phase2/genetic_map/chr${CHROMOSOME}.b38.gmap.gz"
map_impute5=$map_shapeit4
#WGS PATH for accuracy
wgs_subset="/home/ec2-user/adriano/imputation/phase2/WGS/HG00096_example_WGS_data.vcf.gz"
bwgs_subset="/home/ec2-user/adriano/imputation/phase2/WGS/HG00096_example_WGS_data.bcf.gz"

How to run example:

./Imputation_score.sh -i data/HG00096_example_chip_data.vcf.gz -r reference/reference_panel_example_3samples.vcf.gz -t 4 -o chip_example -c 20

Accuracy

Pre-requisite

bcftools
pandas
numpy
cyvcf2

Required command line arguments are:

The following inputs, in vcf.gz format, including its respective tabix .tbi file, are required to run.

• imputed: imputation results
• wgs: ground truth file, containing experimentally determined genotypes (i.e. Whole Genome Sequencing data)
• bwgs: same wgs file but in BCF format to speed up the process and .csi index file associated.

N.B. The Imputation_score.sh is designed to run automatically from phaing to accuracy evaluation, but in case you will need to run accuracy directly on some VCFs file you can use the following section.

How to run Accuracy evaluation:

./bin/imputation_accuracy.sh -i <imputed_phasing_imputation_CombinationSoftware.vcf.gz> -w WGS.vcf.gz -bw WGS.bcf.gz -t 4

This will create multiple chunks of the VCF imputed in input and analyze each one with the Simpy.py software and at the end will use the rebuild_metrics.py script to re-calculate the results.

This message will show up if everything worked:

Imputed VCF   : imputed_HG00479.vcf.gz
WGS VCF       : wgs_HG00479.vcf.gz
WGS BCF       : wgs_HG00479.bcf.gz
Threads       : 4
Skip chunks   : False
Skip analysis : False
Splitted Imputed file in chuncks of [100k]
BCF Imputed files Created
Joining files...
Deleting tmp files...
Process Completed.
${sample_name}_per_sample_results.tsv.gz
${sample_name}_per_variant_results.tsv.gz"

References:

PLOSone: Adriano De Marino, Abdallah Amr Mahmoud, Madhuchanda Bose, Karatuğ Ozan Bircan, Andrew Terpolovsky, Varuna Bamunusinghe, Umar Khan, Biljana Novković, Puya G. Yazdi