rtPRS-CS

rtPRS-CS is a python-based command line tool that performs real-time online updating of polygenic risk score (PRS) weights in a target dataset, one sample at-a-time. Given the most recent set of SNP weights, for each new target sample with both phenotypic and genetic information, rtPRS-CS uses stochastic gradient descent to update the SNP weights, adjusting for the effect of a set of covariates.

Getting Started

• Clone this repository using the following git command:

  git clone https://github.com/getian107/rtPRS.git

  Alternatively, download the source files from the github website (https://github.com/getian107/rtPRS)

• Download the LD reference panels and extract files:

  LD reference panels constructed using the 1000 Genomes Project phase 3 samples:

  AFR reference (~4.44G); tar -zxvf ldblk_1kg_afr.tar.gz

  AMR reference (~3.84G); tar -zxvf ldblk_1kg_amr.tar.gz

  EAS reference (~4.33G); tar -zxvf ldblk_1kg_eas.tar.gz

  EUR reference (~4.56G); tar -zxvf ldblk_1kg_eur.tar.gz

  SAS reference (~5.60G); tar -zxvf ldblk_1kg_sas.tar.gz

  LD reference panels constructed using the UK Biobank data (Notes):

  AFR reference (~4.93G); tar -zxvf ldblk_ukbb_afr.tar.gz

  AMR reference (~4.10G); tar -zxvf ldblk_ukbb_amr.tar.gz

  EAS reference (~5.80G); tar -zxvf ldblk_ukbb_eas.tar.gz

  EUR reference (~6.25G); tar -zxvf ldblk_ukbb_eur.tar.gz

  SAS reference (~7.37G); tar -zxvf ldblk_ukbb_sas.tar.gz

  Note that these files are identical to the reference panels used in PRS-CS.
  Therefore, there is no need to download again if you are already using PRS-CS.

  For regions that don't have access to Dropbox, reference panels can be downloaded from the alternative download site.

• Download the SNP information file and put it in the same folder containing the reference panels:

  1000 Genomes reference: SNP info (~106M)

  UK Biobank reference: SNP info (~108M)

• rtPRS-CS requires Python packages scipy (https://www.scipy.org/) and h5py (https://www.h5py.org/) installed.

• Once Python and its dependencies have been installed, running

  ./rtPRScs.py --help or ./rtPRScs.py -h

  will print a list of command-line options.

Using rtPRS-CS

python rtPRScs.py --ref_dir=PATH_TO_REFERENCE --n_gwas=GWAS_SAMPLE_SIZE --pst_eff=POSTERIOR_EFFECTS --psi_est=PSI_ESTIMATES --vld_prefix=VALIDATION_DATASET_PREFIX --vld_phn_cov=VALIDATION_DATASET_PHENOTYPE_COVARIATES --vld_frq_prefix=VALIDATION_FRQ_FILE_PREFIX --tst_prefix=TESTING_DATASET_PREFIX --tst_phn_cov=TESTING_DATASET_PHENOTYPE_COVARIATES --out_file=OUTPUT_FILENAME [--rate=LEARNING_RATE --imp=IMPLICIT_SGD --order=ORDER_OF_ALGORITHM --tst_phn_cov_update=TESTING_PHENOTYPE_UPDATE --chrom=CHROM]

• PATH_TO_REFERENCE (required): Full path to the directory that contains the SNP information file and LD reference panels. If the 1000 Genomes reference is used, the folder would contain the SNP information file snpinfo_mult_1kg_hm3 and one of the LD reference files: ldblk_1kg_afr, ldblk_1kg_amr, ldblk_1kg_eas, ldblk_1kg_eur, ldblk_1kg_sas; if the UK Biobank reference is used, the folder would contain the SNP information file snpinfo_mult_ukbb_hm3 and one of the LD reference files: ldblk_ukbb_afr, ldblk_ukbb_amr, ldblk_ukbb_eas, ldblk_ukbb_eur, ldblk_ukbb_sas.

• GWAS_SAMPLE_SIZE (required): Sample size of the baseline GWAS.

• POSTERIOR_EFFECTS (required): Starting values for SNP effect sizes, output from running PRS-CS on the baseline GWAS summary statistics.

• PSI_ESTIMATES (required): Shrinkage parameters, output from running PRS-CS (with the option --write_psi) on the baseline GWAS summary statistics.

• VALIDATION_DATASET_PREFIX (required): Full path and the prefix of the bim file for the validation dataset. This file is used to provide a list of SNPs that are available in the validation/target dataset.

• VALIDATION_DATASET_PHENOTYPE_COVARIATES (required): Full path to the space- or tab-delimited text file containing the phenotypes and covariates for the validation samples. The file should follow PLINK's pheno file format (i.e., column 1 contains the family ID, column 2 contains the individual ID, column 3 contains the phenotype, and any remaining columns contain covariates).

• VALIDATION_FRQ_FILE_PREFIX (required): Full path and the prefix of the .frq file output from running PLINK's --freq command on the validation samples.

• TESTING_DATASET_PREFIX (required): Full path and the prefix of the bim file for the testing (target) samples.

• TESTING_DATASET_PHENOTYPE_COVARIATES (required): Full path to the space- or tab-delimited text file containing the phenotypes and covariates for the testing (target) samples. The file should follow PLINK's pheno file format (i.e., column 1 contains the family ID, column 2 contains the individual ID, column 3 contains the phenotype, and any remaining columns contain covariates).

• OUTPUT_FILENAME (required): Output filename prefix of the calculated PRS and final posterior effect size estimates.

• LEARNING_RATE (optional): The learning rate parameter, controlling how each incoming sample is weighted when updating SNP effect sizes. Default is 1.

• IMPLICIT_SGD (optional): A boolean value indicating whether implicit (True) or explicit (False) stochastic gradient descent is used when updating SNP effect sizes. Default is True.

• TESTING_PHENOTYPE_UPDATE (optional): If online phenotype updating is requested, the full path to the space- or tab-delimited text file containing the phenotype, covariates, and position of phenotype update. For subjects requiring phenotype update, column 1 contains the FID, column 2 contains the IID, column 3 contains the original phenotype, column 4 contains the new phenotype, column 5 contains the numeric position in the original phenotype file at which point the target individual's phenotype should be updated, and any remaining columns contain covariates.

• ORDER_OF_ALGORITHM (optional): A string indicating whether the 1st or 2nd order derivative to be used in the stochastic gradient descent algorithm. Default is 2nd.

• CHROM (optional): A numeric value or range of values indicating which chromosome(s) should be run. Default is iterating through 22 autosomes.

Output

rtPRS-CS writes two output files to the user-specified directory. The first is a .txt file containing the standardized PRS for each target sample. The second is a .pst_eff file containing the vector of SNP weights after incorporating all target samples.

Support

Please direct any problems or questions to Justin Tubbs (jtubbs@mgh.harvard.edu) or Tian Ge (tge1@mgh.harvard.edu).

Toy Example

Simulated data for chromosome 22 is provided in /toy_data/ to demonstrate the usage of rtPRS-CS. It assumes that a copy of the EUR LD reference panel has been downloaded (see above).

1. The first step is to run PRS-CS-auto on the baseline summary statistics to obtain starting estimates of the SNP weights (baseline_pst_eff_a1_b0.5_phiauto_chr22.txt), along with estimates of psi (baseline_pst_psi_a1_b0.5_phiauto_chr22.txt).

python PRScs.py --ref_dir=/path/to/ldblk_ukbb_eur --bim_prefix=./toy_data/1000GP_HM3 --sst_file=./toy_data/toy_sumstats.txt --n_gwas=50000 --write_psi=True --chrom=22 --out_dir=./toy_data/baseline

2. rtPRS-CS can be run using the output from (1) along with a validation and test dataset. The output will be a PRS for each sample in the test dataset (toy_out_prs_rate1.0_imp_2nd_chr22.txt), along with the final set of SNP weights after incorporating all test samples (toy_out_psteff_rate1.0_imp_2nd_chr22.txt).

python rtPRScs.py --ref_dir=/path/to/ldblk_ukbb_eur --n_gwas=50000 --pst_eff=./toy_data/baseline_pst_eff_a1_b0.5_phiauto_chr22.txt --psi_est=./toy_data/baseline_pst_psi_a1_b0.5_phiauto_chr22.txt --vld_prefix=./toy_data/1000GP_HM3_EUR_chr22_vld --vld_phn_cov=./toy_data/pheno_vld.txt --tst_prefix=./toy_data/1000GP_HM3_EUR_chr22_tst --tst_phn_cov=./toy_data/pheno_tst.txt --chrom=22 --out_file=./toy_data/toy_out