michiganCoxSurv: Fit Cox survival to all variants in a .vcf.gz file from...

Description Usage Arguments Details Value Examples

View source: R/michiganCoxSurv.R

Description

Performs survival analysis using Cox proportional hazard models on imputed genetic data stored in compressed VCF files

Usage

1
2
3
4
michiganCoxSurv(vcf.file, covariate.file, id.column, sample.ids = NULL,
  time.to.event, event, covariates, inter.term = NULL,
  print.covs = "only", out.file, maf.filter = 0.05, r2.filter = NULL,
  chunk.size = 5000, verbose = TRUE, clusterObj = NULL)

Arguments

vcf.file

character(1) path to VCF file.

covariate.file

matrix(1) comprising phenotype (time, event) and additional covariate data.

id.column

character(1) providing exact match to sample ID column from covariate.file

sample.ids

character vector with sample ids to include in analysis

time.to.event

character(1) string that matches time column name in pheno.file

event

character(1) string that matches event column name in pheno.file

covariates

character vector with matching column names in pheno.file of covariates of interest

inter.term

character(1) string giving the column name of the covariate that will be added to the interaction term with SNP (e.g. term*SNP). See details.

print.covs

character(1) string of either "only", "all" or "some", defining which covariate statistics should be printed to the output. See details.

out.file

character(1) string with output name

maf.filter

integer(1) filter out minor allele frequency below threshold (i.e. 0.005 will filter MAF > 0.005)

r2.filter

integer(1) of imputation quality score filter (i.e. 0.7 will filter r2 > 0.7)

chunk.size

integer(1) number of variants to process per thread

verbose

logical(1) for messages that describe which part of the analysis is currently being run

clusterObj

A cluster object that can be used with the parApply function. See details.

Details

Testing for SNP-covariate interactions: User can define the column name of the covariate that will be included in the interaction term. For example, for given covariates a and b, where c is defined as the inter.term the model will be: ~ a + b + c + SNP + c*SNP.

Printing results of other covariates: print.covs argument controls the number of covariates will be printed as output. The function is set to only by default and will only print the SNP or if an interaction term is given, the results of the interaction term (e.g. SNP*covariate). Whereas, all will print results (coef, se.coef, p.value etc) of all covariates included in the model. some is only applicable if an interaction term is given and will print the results for SNP, covariate tested for interaction and the interaction term. User should be mindful about using the all option, as it will likely slow down the analysis and will increase the output file size.

User defined parallelization: This function uses parApply from parallel package to fit models to SNPs in parallel. User is not required to set any options for the parallelization. However, advanced users who wish to optimize it, can provide a cluster object generated by makeCluster family of functions that suits their need and platform.

Value

Saves two text files directly to disk: .coxph extension containing CoxPH survival analysis results. .snps_removed extension containing SNPs that were removed due to low variance or user-defined thresholds.

Examples

 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
vcf.file <- system.file(package="gwasurvivr",
                        "extdata",
                     "michigan.chr14.dose.vcf.gz")
pheno.fl <- system.file(package="gwasurvivr",
                        "extdata",
                     "simulated_pheno.txt")
pheno.file <- read.table(pheno.fl, 
                         sep=" ",
                         header=TRUE,
                         stringsAsFactors = FALSE)
pheno.file$SexFemale <- ifelse(pheno.file$sex=="female", 1L, 0L)
sample.ids <- pheno.file[pheno.file$group=="experimental",]$ID_2
michiganCoxSurv(vcf.file=vcf.file,
              covariate.file=pheno.file,
              id.column="ID_2",
              sample.ids=sample.ids,
              time.to.event="time",
              event="event",
              covariates=c("age", "SexFemale", "DrugTxYes"),
              inter.term=NULL,
              print.covs="only",
              out.file="michigan_example",
              r2.filter=0.3,
              maf.filter=0.005,
              chunk.size=50,
              verbose=TRUE,
              clusterObj=NULL)

gwasurvivr documentation built on Nov. 8, 2020, 6:53 p.m.