#cp the data in /home/ubuntu/Day1 folder and its content in my home
cp -r Share/Day1/ .

cd Day1/
ls -l

#take a look at the files
head data/training_tobeimputed.map #chr snp_id position
cut -d' ' -f1-10 data/training_tobeimputed.ped |head #Family ID, Individual ID, Paternal ID, Maternal ID , Sex , Phenotype, SNP1_REF, SNP1_ALT, SNP2_REF, SNP2_REF...

#Transforming .ped and .map files into vcf (required by Beagle)
plink  --file data/training_tobeimputed --recode vcf --out data/training

#List of the files that were generated
ls -l data

# training.log and training.vcf files were generated. Check the log file for any warning
#inspect vcf file 
head -12 data/training.vcf #Lin 10: #CHROM  POS     ID      REF     ALT     QUAL    FILTER  INFO    FORMAT  9_sample1 ...


#Running Beagle 5.1 to phase the genotypes in the reference panel
java -Xss5m -Xmx4g -jar beagle.18May20.d20.jar gt=data/training.vcf out=data/training_ref
#list the files
ls -l data
# Two files were generated: training_ref.vcf.gz and training_ref.log
#Inspecting the first lines of the phased vcf file 
zcat data/training_ref.vcf.gz |head -12

#Transforming .ped and .map files of the target population into vcf (required by Beagle)
plink  --file data/testing_tobeimputed --recode vcf --out data/testing_tobeimputed

#List of the files that were generated
ls -l data

#Two files were generated testing_tobeimputed.vcf and testing_tobeimputed.log


#Running Beagle 5.1 to impute missing genotypes in the target population using the reference panel
java -Xss5m -Xmx4g -jar  beagle.18May20.d20.jar  gt=data/testing_tobeimputed.vcf ref=data/training_ref.vcf.gz out=data/testing_imputed

#List of the files that were generated
ls -l data

#Two files were generated testing_imputed.vcf.gz and testing_imputed.log

#Inspecting the first lines of the phased vcf file 
zcat data/testing_imputed.vcf.gz |head -12

#transforming the vcf file (output from Beagle) into a dosage file
plink --vcf data/testing_imputed.vcf.gz --recode A --out data/testing_imputed_dosage

#
ls -l data
#three files were generated:  testing_imputed_dosage.log, testing_imputed_dosage.nosex, testing_imputed_dosage.raw (the dosage file)