build_predictor: Uses output from filter_regions() to build the predictor

Description Usage Arguments Value Examples

View source: R/build_predictor.R

Description

This function takes the output from filter_regions() and builds the phenotype predictor to be used for phenotype prediction in a new data set. This function outputs the fits (coefficient estimates, 'coefEsts') from the model for the selectied regions as well as the rows selected ('trainingProbes') from the input data.

Usage

1
2
build_predictor(inputdata = NULL, phenodata = NULL, phenotype = NULL,
  covariates = NULL, type = NULL, numRegions = NULL)

Arguments

inputdata

output from filter_regions() inputdata

phenodata

data set with phenotype information; samples in rows, variables in columns phenodata

phenotype

phenotype of interest phenotype

covariates

Which covariates to include in model covariates

type

The class of the phenotype of interest (numeric, factor) type

numRegions

The number of regions per class of variable of interest to pull out from each chromosome (default: 10) numRegions

Value

An n x m data.frame of coefficient estimates and region indices for each of the regions included from filter_regions() along with regiondata and indices for trainingProbes

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
library('GenomicRanges')
library('dplyr')

## Make up some some region data
regions <- GRanges(seqnames = 'chr2', IRanges(
     start = c(28971710:28971712, 29555081:29555083, 29754982:29754984),
     end = c(29462417:29462419, 29923338:29923340, 29917714:29917716)))

## make up some expression data for 9 rows and 30 people
data(sysdata, package='phenopredict')
## includes R object 'cm'
exp= cm[1:length(regions),1:30]

## generate some phenotype information
sex = as.data.frame(rep(c("male","female"),each=15))
age = as.data.frame(sample(1:100,30))
pheno = dplyr::bind_cols(sex,age)
colnames(pheno) <- c("sex","age")

## filter regions to be used to build the predictor
inputdata <- filter_regions(expression=exp, regiondata=regions,
phenodata=pheno, phenotype="sex", covariates=NULL,type="factor",
numRegions=2)

## build phenotype predictor
predictor<-build_predictor(inputdata=inputdata ,phenodata=pheno,
	phenotype="sex", covariates=NULL,type="factor", numRegions=2)

leekgroup/phenopredict documentation built on May 14, 2019, 11:27 a.m.