Ksea

Introduction

This package contains an implementation of the Kinase Set Enrichment Analysis (KSEA) used to predict kinase activities based on quantitative phosphoproteomic studies. Using a quantitative phosphoproteomic profile and a list of known targets, the algorithms predicts the activity of the based on the enrichment on top regulated sites within the known targets of the kinase.

Installation

The ksea package can be installed directly from github (if public) or locally using the devtools package.

install.packages('devtools')

Github installation

require(devtools)
install_github("evocellnet/ksea")

Local installation

You can clone this project and install it locally in your computer.

require(devtools)
install("./ksea")

Usage

First load the ksea package.

library("ksea")

Next create some fake data. In one hand, we create a list named regulons containing a vector per kinase with the names of the known substrates. Secondly, we create a vector sites with quantifications for the sites going from A to Z. Finally, we sort the quantifications.

regulons <- list(kinaseA=sample(LETTERS, 5))
regulons

## $kinaseA
## [1] "Y" "C" "I" "S" "B"

sites <- rnorm(length(LETTERS))
names(sites) <- LETTERS
sites <- sites[order(sites, decreasing=TRUE)]
sites

##            E            U            Z            N            S 
##  1.611089290  1.250664569  1.234637179  1.082452371  1.026981805 
##            B            Q            J            M            D 
##  0.575379472  0.532877556  0.474246843  0.421483820  0.358049788 
##            F            K            Y            G            O 
##  0.194403791  0.162551646  0.150628008  0.008759558 -0.150678120 
##            P            C            W            V            T 
## -0.156401856 -0.237702939 -0.240834542 -0.254118235 -0.317312680 
##            H            R            A            I            X 
## -0.337654466 -0.406046178 -0.433621274 -0.585813866 -1.189626572 
##            L 
## -1.706905250

The function ksea will run the enrichment analysis for the provided quantifications and known kinase targets.

ksea_result <- ksea(names(sites), sites, regulons[["kinaseA"]], trial=1000, significance = TRUE)

ksea_result

## $ES
##         B 
## 0.4314363 
## 
## $p.value
##     B 
## 0.447

The function ksea_batchKinases calculates the KSEA p-value for a list of kinases. To improve the performance of the function, it uses as many cores as possible using the parallell package.

regulons[["kinaseB"]] <- sample(LETTERS, 3)
regulons[["kinaseC"]] <- sample(LETTERS, 7)
regulons

## $kinaseA
## [1] "Y" "C" "I" "S" "B"
## 
## $kinaseB
## [1] "J" "N" "E"
## 
## $kinaseC
## [1] "W" "E" "C" "Z" "G" "A" "U"

kinases_ksea <- ksea_batchKinases(names(sites), sites, regulons, trial=1000)

## Loading required package: parallel

kinases_ksea

## kinaseA.B kinaseB.J kinaseC.Z 
##     0.439     0.091     0.005

KSEA in parallel

Some functions such as ksea_batchkinases are optimized to run in parallel in multicore processors. By default they run in 2 cores (if available) but this number can be increased by changing the "mc.cores" variable.

options("mc.cores" = 4)

If you are working with an LSF cluster, you can also take the number of cores allocated in the bsub command from the LSB_MCPU_HOSTS environment variable.

## Setup multicore forking for mclapply based in bsub available cores
hosts <- Sys.getenv("LSB_MCPU_HOSTS")
if(length(hosts) >0){
ncores <- unlist(strsplit(hosts," "))[2]
}else{
ncores <- 1
}
options("mc.cores" = ncores)

Developers

The package is documented using roxygen2. After changing the code located in the R/ folder remember to run 'make' on the main directory to create the documentation following the roxygen2 rules.

To work on the code you will need the devtools package installed (installation guide above).