Biobank-Scale Simulator

Overview

This repository contains tools for simulating phenotypes at biobank scale, incorporating various genetic and environmental components.

Simulation Types

1. Phenotypes with Additive (G) and Noise Components:

   • Simulator_annot: Simulates phenotypes based on SNP annotations and their variance components.
   • Simulator_mafld: Simulates phenotypes based on MAF and LD-dependent genetic architecture.

2. Phenotypes with Additive (G), Gene-by-Environment (GxE), Noise-by-Environment (NxE), and Noise Components:

   • Simulator_gxe: Simulates phenotypes with G + GxE + NxE components (or combinations as specified in the parameter file).
   • Simulator_gxe_with_corr: Simulates phenotypes correlated with the environment.
   • Simulator_gxe_same_causal: Simulates phenotypes with the same causal SNPs for G and GxE components.
   • Simulator_gxe_causal_from_file: Simulates phenotypes with user-specified causal SNPs for the G component.
   • Simulator_gxe_heavy_tail: Simulates phenotypes with noise from a heavy-tailed distribution (Student's t-distribution with DoF = 4).
   • Simulator_gxe_multi_E: Simulates phenotype with GxE effects from different environments.

3. Phenotypes with Additive (G), Gene-by-gene (GxG), and Noise Components:

   • Simulator_gxg: Simulates phenotypes with G + GxG components.

---

Prerequisites

The following packages are required on a Linux machine to compile and use this software:

g++
cmake
make

Installation

To set up the simulator, clone the repository and build the project:

git clone https://github.com/sriramlab/Simulator.git
cd Simulator
mkdir build
cd build/
cmake ..
make

Inputs

Simulator_annot

Parameters:

• -g: Genotype file in BED format
• -annot: Annotation file with M+1 rows and K columns (M = number of SNPs, K = number of annotations).
  • SNP annotations are binary (1 = belongs to annotation, 0 = otherwise).
  • The first line specifies the true variance components.
• -o: Output directory
• -k: Number of phenotypes
• -jn: Number of stream blocks (to reduce memory usage).

Simulator_mafld

Parameters:

• -g: Genotype file
• -annot: Annotation file
  • M rows (M = number of SNPs), 1 column (1 = included, 0 = not included).
• -o: Output directory
• -maf_ld: File with MAF and LD of SNPs.
• -simul_par: Simulation parameters file with 7 columns.
  • Seven parameters: percentage of causal SNPs, exponent of LD,exponent of MAF, min(MAF) of causal SNPs, max(MAF) of causal SNPs, total h2, number of simulations.
  • Refer to this publication for details.
• -jn: Number of stream blocks (to reduce memory usage).

Simulator_gxe

Parameters:

• -g: Genotype file in BED format
• -e: Environment file in TXT format
  • N rows (N = number of individuals), L columns (L = number of environments).
• -annot: Annotation file in TXT format
• -o: Output directory
• -maf_ld: File with MAF and LD of SNPs
• -simul_par: Simulator parameter file with 3 lines (for G, GxE, and NxE components).
• -jn: Number of stream blocks

Simulator_gxe_with_corr

Parameters:

• Same as Simulator_gxe plus:
  • -corr_w_E: Correlation with the environment (E).

Simulator_gxe_same_causal

Parameters:

• Same as Simulator_gxe plus:
  • -causal_snp_file: Output file specifying the causal SNPs (1 = causal, 0 = not causal).

Simulator_gxe_same_causal

Parameters:

• Same as Simulator_gxe plus:
  • -causal_snp_file: Input file specifying causal SNPs for the additive (G) component.

Simulator_gxe_heavy_tail

Parameters:

• Same as Simulator_gxe

Simulator_gxe_multi_E

Parameters:

• Same as Simulator_gxe
• -simul_par: Simulator parameter file with N_env+2 lines (for G, GxE, and NxE components). The 2nd to N_env+1 lines are for GxE component under different environments. Each environment corresponds to a separate model with G+GxE+NxE.

Simulator_gxg

Parameters:

• -g : Genotype file in BED format
• -annot: Annotation file in TXT format
  • M rows (M=number of SNPs) and 3 columns.
  • First column corresponds to SNPs with additive effect.
  • Second and third column correspond to two groups of SNPs which have interaction with each other (there is no interaction inside each group). Third column/group is assumed to be smaller than second.
  • If SNP i belongs to annotation j, then there is "1" in row i and column j. Otherwise there is "0". (delimiter is " ")
• -jn: The number of stream blocks. (the higher number of blocks, the lower memory usage)
• -o: The path of output file.

Examples

Toy examples are provided in the example folder:

• test.sh
• test.gxe.sh
• test.gxg.sh

Citations

1. Ali Pazokitoroudi, Zhengtong Liu, Andrew Dahl, Noah Zaitlen, Saharon Rosset, Sriram Sankararaman. AJHG (2024); DOI: 10.1016/j.ajhg.2024.05.015

2. Ali Pazokitoroudi, Yue Wu, Kathryn S. Burch, Kangcheng Hou, Aaron Zhou, Bogdan Pasaniuc, Sriram Sankararaman. Nature Communications (2020); DOI: 10.1038/s41467-020-17576-9

3. Boyang Fu*, Ali Pazokitoroudi*, Albert Xue, Aakarsh Anand, Prateek Anand, Noah Zaitlen, Sriram Sankararaman. bioRxiv, 2020. DOI: 10.1101/2023.09.10.557084