TagSNP_evaluation

Evaluation tag SNP selection strategies

These scipts tested in Ubuntu v18 environment.

Software requirements

Leave one out imputation cross validation requires R, vcftools, bcftools, minimac3, minimac4, and plink.

• VCFtools 0.1.17 : https://vcftools.github.io
• bcftools 1.10.2 : https://github.com/samtools/bcftools
• plink1.9 : https://www.cog-genomics.org/plink/
• minimac3 : https://genome.sph.umich.edu/wiki/Minimac3
• minimac4 : https://genome.sph.umich.edu/wiki/Minimac4
• R packages version 3.6.0 or latter

Clone this project and config pipeline:

git clone https://github.com/datngu/TagSNP_evaluation

I. RUNNING PIPELINES

Assumming that you have a processed (Bialellic SNPs with MAF >= 1%) vcf.gz file named: chr10_EAS.vcf.gz and want to compare impuatation accuracy with 30000 tag SNP selected.

1. Running TagIt

cd TagSNP_evaluation/TagIt
bash configure.sh

bash TagIt_pipeline.sh ../chr10_EAS.vcf.gz chr10_EAS

# picking top 30000 tag SNP selected from 
head -n 30000 chr10_EAS/chr10_EAS_tags_cleaned.txt > chr10_EAS_tags_30000_cleaned.txt

You output should be: TagSNP_evaluation/TagIt/chr10_EAS_tags_30000_cleaned.txt

2. Running FastTagger

cd TagSNP_evaluation/FastTagger 

bash bash configure.sh

bash FastTagger_pipeline.sh  ../chr10_EAS.vcf.gz chr10_EAS 30000 chr10_EAS

You output should be: TagSNP_evaluation/FastTagger/cleaned_EAS_chr10_EAS_30000.tagSNP.txt

3. Running EQ_uniform

cd TagSNP_evaluation/EQ_uniform

Rscript EQ_uniform.R vcf=../chr10_EAS.vcf.gz size=30000 out=EQ_uniform_array_30000.txt

You output should be: TagSNP_evaluation/EQ_uniform/EQ_uniform_array_30000.txt

4. Running EQ_MAF

cd TagSNP_evaluation/EQ_MAF

Rscript EQ_MAF.R vcf=../chr10_EAS.vcf.gz size=30000 out=EQ_MAF_array_30000.txt

You output should be: TagSNP_evaluation/EQ_MAF/EQ_MAF_array_30000.txt

II. EVALUATION WITH LEAVE ONE OUT CROSS VALIDATION

1. Create imputation reference pannel

This step take very long time, depends on your size of chromosome and population.

cd TagSNP_evaluation
create_imputation_ref.sh -v chr10_EAS.vcf.gz -o chr10_EAS_imputation_ref -p 16

2. Leave one out imputation with your pre-buit reference panel and computing Pearson's correlation

cd TagSNP_evaluation
# 1. TagIt
## imputation
imputation_with_prebuilt_ref.sh -t TagIt/chr10_EAS_tags_30000_cleaned.txt -r chr10_EAS_imputation_ref -o TagIt_EAS -p 16
## computing Pearson's correlation
compute_imputation_accuracy.R imputation=TagIt_EAS out=TagIt_EAS.Rdata

# 2. FastTagger
## imputation
imputation_with_prebuilt_ref.sh -t FastTagger/cleaned_EAS_chr10_EAS_30000.tagSNP.txt -r chr10_EAS_imputation_ref -o FastTagger_EAS -p 16
## computing Pearson's correlation
compute_imputation_accuracy.R imputation=FastTagger_EAS out=FastTagger_EAS.Rdata


# 3. EQ_uniform
## imputation
imputation_with_prebuilt_ref.sh -t EQ_uniform/EQ_uniform_array_30000.txt -r chr10_EAS_imputation_ref -o EQ_uniform_EAS -p 16
## computing Pearson's correlation
compute_imputation_accuracy.R imputation=EQ_uniform_EAS out=EQ_uniform_EAS.Rdata

# 4. EQ_MAF
## imputation
imputation_with_prebuilt_ref.sh -t EQ_MAF/EQ_MAF_array_30000.txt -r chr10_EAS_imputation_ref -o EQ_MAF_EAS -p 16
## computing Pearson's correlation
compute_imputation_accuracy.R imputation=EQ_MAF_EAS out=EQ_MAF_EAS.Rdata

You output should be: TagIt_EAS.Rdata, FastTagger_EAS.Rdata, EQ_uniform_EAS.Rdata, EQ_MAF_EAS.Rdata

III. REFERENCE

Nguyen, D. T., Dinh, H. Q., Vu, G. M., Nguyen, D. T., & Vo, N. S. (2021, November). A comprehensive imputation-based evaluation of tag SNP selection strategies. In 2021 13th International Conference on Knowledge and Systems Engineering (KSE) (pp. 1-6). IEEE. https://doi.org/10.1109/KSE53942.2021.9648614