In Osmahmoud/SlopeHunter: Slope-Hunter: A ROBUST METHOD FOR COLLIDER BIAS CORRECTION IN CONDITIONAL GENOME-WIDE ASSOCIATION STUDIES

Software Installation

Requirements

The SlopeHunter software does not require any special hardware. A standard computer suitable for most applications nowadays should be sufficient.

SlopeHunter is written as an R package that requires R (>= 3.5.0) to be installed on your machine. If you do NOT have R installed, or you do have an older version of R installed on your machine, you would need to install the latest R version from here.

The SlopeHunter R package depends on a few other R packages that will be automatically installed with SlopeHunter when you follow the installation guide below. A list of these R dependencies is reported here.

System Compatibility

The SlopeHunter software is compatible with Windows, Linux and macOS. The package has been tested in R under the following systems:

• Linux: Ubuntu 16.04 (R 4.0.4)
• macOS: Monterey 12.3 (R 4.2)
• Windows: 10 (R 4.2)

Installation Guide

To install the SlopeHunter R package, simply run

# required only once per machine!
devtools::install_github("Osmahmoud/SlopeHunter")

Reading Data

We are going to use data on ... (GIVE DESCRIPTION on BCa DATA)

First, you need to load the SlopeHunter package:

require(SlopeHunter)

Reading in the incidence GWAS

BCa_incidence <- SlopeHunter::read_incidence(filename = "Your-incidence-data-file.gz",
  )

Install the SlopeHunter R package by following the installation instructions described here.

Slope-Hunter

Report tutorials on the simulated data here ...

Applied Data Analysis

Example: Fasting Insulin adjusted for BMI

Format Data for Slope-Hunter Analysis

First, you need to load the Slope-Hunter R package:

library(SlopeHunter)

knitr::opts_chunk$set(eval = FALSE)

Read in the incidence (the condition trait) data

BMI_incidence <- read_incidence(filename = "Your-incidence-data-file.gz",
  gz = TRUE,
  eaf_col="Freq_Tested_Allele_in_HRS",
  effect_allele_col="Tested_Allele",
  other_allele_col="Other_Allele",
  pval_col = "P",
  pos_col = "POS")

Read in the conditional outcome data (BMI-adjusted Insulin)

Insulin_adj_BMI <- read_prognosis("ftp://ftp.ebi.ac.uk/pub/databases/gwas/summary_statistics/MahajanA_25625282_GCST007857/Mahajan_2014_SV_ExomeChip_lnFI_AdjForBMI.txt", 
                                  gz = FALSE,
                                  sep = "\t",
                                  snp_col = "ID",
                                  beta_col = "BETA",
                                  se_col = "SE",
                                  pval_col = "P",
                                  effect_allele_col = "EA",
                                  other_allele_col = "NEA",
                                  chr_col = "Chromosome",
                                  pos_col = "Position")

Insulin_adj_BMI <- read_prognosis("Your-conditional-outcome-data-file.txt", 
                                  gz = FALSE,
                                  sep = "\t",
                                  snp_col = "ID",
                                  beta_col = "BETA",
                                  se_col = "SE",
                                  pval_col = "P",
                                  effect_allele_col = "EA",
                                  other_allele_col = "NEA",
                                  chr_col = "Chromosome",
                                  pos_col = "Position")

Harmonisation

Data_harmonised <- harmonise_effects(incidence_dat = BMI_incidence,
                                prognosis_dat = Insulin_adj_BMI,
                                by.pos = TRUE, pos_cols = c("POS.incidence", "POS.prognosis"),
                                snp_cols=c("SNP", "SNP"),
                                beta_cols = c("BETA.incidence", "BETA.prognosis"),
                                se_cols=c("SE.incidence", "SE.prognosis"),
                                EA_cols=c("EA.incidence", "EA.prognosis"),
                                OA_cols=c("OA.incidence", "OA.prognosis")
                                )

nrow(Data_harmonised)               # No. SNPs present in both datasets
attributes(Data_harmonised)$info    # Get info on the harmonisation process

Extract valid data after harmonisation (those that are eligible for pruning)

Data_to_prune <- Data_harmonised[!Data_harmonised$remove, ]
nrow(Data_to_prune)                 # No. SNPs to be pruned

Pruning data

Data_pruned <- LD_prune(Data_to_prune, Random = TRUE, seed = 15151515)

Apply the Slope-Hunter approach

TBA ...

Osmahmoud/SlopeHunter documentation built on Oct. 7, 2022, 4:38 p.m.