| Title: | Signaling Pathway Impact Analysis (SPIA) using combined evidence of pathway over-representation and unusual signaling perturbations |
|---|---|
| Description: | This package implements the Signaling Pathway Impact Analysis (SPIA) which uses the information form a list of differentially expressed genes and their log fold changes together with signaling pathways topology, in order to identify the pathways most relevant to the condition under the study. |
| Authors: | Adi Laurentiu Tarca <[email protected]>, Purvesh Kathri <[email protected]> and Sorin Draghici <[email protected]> |
| Maintainer: | Adi Laurentiu Tarca <[email protected]> |
| License: | file LICENSE |
| Version: | 2.59.0 |
| Built: | 2024-10-31 05:29:00 UTC |
| Source: | https://github.com/bioc/SPIA |
The colorectal dataset consists: i) an named vector DE_Colorectal
, which represents the $log2$ fold changes of the genes chosen as differentially expressed between
colorectal cancer and normal samples based on data from Hong et al, 2007, using a $FDR=0.1$ and
the universe of all Entrez gene IDs available on the array, ALL_Colorectal. These two vectors were obtained starting from the
top dataframe which is the output from the topTable function of the limma package using the RMA processed gene expression data downloaded from
GEE (GSE4107).
The microarray platform used was Affymetrix HGU-133PLUS2.0.
data(colorectalcancer)data(colorectalcancer)
Yi Hong and Kok Sun Ho and Kong Weng Eu and Peh Yean Cheah, A susceptibility gene set for early onset colorectal cancer that integrates diverse signaling pathways: implication for tumorigenesis, Clin Cancer Res, 2007, 13(4),1107-14.
Combining two p-values using Fisher's product or normal inversion methods.
combfunc(p1=NULL,p2=NULL,combine="fisher")combfunc(p1=NULL,p2=NULL,combine="fisher")
p1 |
A vector of probabilities. |
p2 |
A vector of probabilities. |
combine |
A string with the name of the method to be used. Options include "fisher","norminv" |
Two vectors of p-values are combined into a vector of global p-values.
A vector of p-values.
Adi Laurentiu Tarca <[email protected]>, Purvesh Khatri, Sorin Draghici
Adi L. Tarca, Sorin Draghici, Purvesh Khatri, et. al, A Signaling Pathway Impact Analysis for
Microarray Experiments, 2008, Bioinformatics, 2009, 25(1):75-82.
# Examples use colorectal cancer dataset p1=c(0.2,0.4,0.1) p2=c(0.01,0.7,0.01) pG=combfunc(p1,p2,combine="fisher") pG=combfunc(p1,p2,combine="norminv")# Examples use colorectal cancer dataset p1=c(0.2,0.4,0.1) p2=c(0.01,0.7,0.01) pG=combfunc(p1,p2,combine="fisher") pG=combfunc(p1,p2,combine="norminv")
This function processes KEGG xml files into a xxxSPIA.RData file needed for spia function.
makeSPIAdata(kgml.path="./hsa",organism="hsa",out.path=".")makeSPIAdata(kgml.path="./hsa",organism="hsa",out.path=".")
kgml.path |
Character vector giving the location of the folder containing two or more KEGG xml files. See for e.g. http://www.genome.jp/kegg/pathway/hsa/hsa04010.html and click the Download KGML to get such files. Users that have a license to the KEGG ftp directory can copy all the xml files corresponding to a givne organism. |
organism |
A three letter character designating the organism. See a full list at ftp://ftp.genome.jp/pub/kegg/xml/organisms. |
out.path |
Directory where a "organism"SPIA.RData file will be saved. If left to null, it will will try to save the file in th extdata folder of the SPIA library. |
Adi Laurentiu Tarca <[email protected]>
library(SPIA) data(colorectalcancer) makeSPIAdata(kgml.path=system.file("extdata/keggxml/hsa",package="SPIA"),organism="hsa",out.path="./") res<-spia(de=DE_Colorectal, all=ALL_Colorectal, organism="hsa",data.dir="./") res[,-12]library(SPIA) data(colorectalcancer) makeSPIAdata(kgml.path=system.file("extdata/keggxml/hsa",package="SPIA"),organism="hsa",out.path="./") res<-spia(de=DE_Colorectal, all=ALL_Colorectal, organism="hsa",data.dir="./") res[,-12]
Plots each pathway as a point, using the over-representation p-value, pNDE, and perturbations accumulation p-value, pPERT, as coordinates. In addition the regions where FDR and FWER adjusted pG values are less than the specified threshold are plotted. The function determines automatically which method (fisher or norminv) was used to combine the two p-values into pG, and plots the regions described above accordingly.
plotP(x,threshold=0.05)plotP(x,threshold=0.05)
x |
A data frame produced by |
threshold |
A numerical value between 0 and 1 to be used as significance threshold in inferring pathway significance. |
In this plot each pathway is a point and the coordinates are the log of pNDE (using a hypergeometric model) and the p-value from perturbations, pPERT. The oblique lines in the plot show the significance regions based on the combined evidence.
This function does not return any value. It only generates a plot.
Adi Laurentiu Tarca <[email protected]>, Purvesh Khatri, Sorin Draghici
Adi L. Tarca, Sorin Draghici, Purvesh Khatri, et. al, A Signaling Pathway Impact Analysis for
Microarray Experiments, 2008, Bioinformatics, 2009, 25(1):75-82.
# Examples use colorectal cancer dataset data(colorectalcancer) # pathway analysis based on combined evidence of ORA and perturbations # use nB=2000 or larger for more accurate results res<-spia(de=DE_Colorectal, all=ALL_Colorectal, organism="hsa",nB=200,plots=FALSE,verbose=TRUE,beta=NULL,combine="fisher") #Generate the evidence plot plotP(res,threshold=0.1) res<-spia(de=DE_Colorectal, all=ALL_Colorectal, organism="hsa",nB=200,plots=FALSE,verbose=TRUE,beta=NULL,combine="norminv") #Generate the evidence plot plotP(res,threshold=0.1)# Examples use colorectal cancer dataset data(colorectalcancer) # pathway analysis based on combined evidence of ORA and perturbations # use nB=2000 or larger for more accurate results res<-spia(de=DE_Colorectal, all=ALL_Colorectal, organism="hsa",nB=200,plots=FALSE,verbose=TRUE,beta=NULL,combine="fisher") #Generate the evidence plot plotP(res,threshold=0.1) res<-spia(de=DE_Colorectal, all=ALL_Colorectal, organism="hsa",nB=200,plots=FALSE,verbose=TRUE,beta=NULL,combine="norminv") #Generate the evidence plot plotP(res,threshold=0.1)
This function implements the SPIA algorithm to analyze KEGG signaling pathways.
spia(de=NULL,all=NULL,organism="hsa",data.dir=NULL,pathids=NULL,nB=2000,plots=FALSE,verbose=TRUE,beta=NULL,combine="fisher")spia(de=NULL,all=NULL,organism="hsa",data.dir=NULL,pathids=NULL,nB=2000,plots=FALSE,verbose=TRUE,beta=NULL,combine="fisher")
de |
A named vector containing log2 fold-changes of the differentially expressed genes. The names of this numeric vector are Entrez gene IDs. |
all |
A vector with the Entrez IDs in the reference set. If the data was obtained from a microarray experiment,
this set will contain all genes present on the specific array used for the experiment. This vector should
contain all names of the |
organism |
A three letter character designating the organism. See a full list at ftp://ftp.genome.jp/pub/kegg/xml/organisms. |
data.dir |
Location of the "organism"SPIA.RData file containing the pathways data generated with makeSPIAdata. If set to NULL will look for this file in the extdata folder of the SPIA library. |
pathids |
A character vector with the names of the pathways to be analyzed. If left NULL all pathways available will be tested. |
nB |
Number of bootstrap iterations used to compute the P PERT value. Should be larger than 100. A recommended value is 2000. |
plots |
If set to TRUE, the function plots the gene perturbation accumulation vs log2 fold change for every gene on each pathway. The null distribution of the total net accumulations from which PPERT is computed, is plotted as well. The figures are sent to the SPIAPerturbationPlots.pdf file in the current directory. |
verbose |
If set to TRUE, displays the number of pathways already analyzed. |
beta |
Weights to be assigned to each type of gene/protein relation type. It should be a named numeric vector of length 23, whose names must be:
If set to null, beta will be by default chosen as: c(1,0,0,1,-1,1,0,0,-1,-1,0,0,1,0,1,-1,0,1,-1,-1,0,0,0). |
combine |
Method used to combine the two types of p-values. If set to |
See cited documents for more details.
A data frame containing the ranked pathways and various statistics: pSize is the number of genes on the pathway;
NDE is the number of DE genes per pathway; tA is the observed total preturbation
accumulation in the pathway; pNDE is the probability to observe at least NDE genes on
the pathway using a hypergeometric model;
pPERT is the probability to observe a total accumulation more extreme than tA only by
chance;
pG is the p-value obtained by combining pNDE and pPERT;
pGFdr and pGFWER are the
False Discovery Rate and respectively Bonferroni adjusted global p-values; and the Status gives the direction
in which the pathway is perturbed (activated or inhibited).
KEGGLINK gives a web link to the KEGG website that displays the pathway image with the differentially expressed genes
highlighted in red.
Adi Laurentiu Tarca <[email protected]>, Purvesh Khatri, Sorin Draghici
Adi L. Tarca, Sorin Draghici, Purvesh Khatri, et. al, A Signaling Pathway Impact Analysis for
Microarray Experiments, 2008, Bioinformatics, 2009, 25(1):75-82.
Purvesh Khatri, Sorin Draghici, Adi L. Tarca, Sonia S. Hassan, Roberto Romero. A system biology
approach for the steady-state analysis of gene signaling networks. Progress in Pattern Recognition,
Image Analysis and Applications, Lecture Notes in Computer Science. 4756:32-41, November 2007.
Draghici, S., Khatri, P., Tarca, A.L., Amin, K., Done, A., Voichita, C., Georgescu, C., Romero, R.:
A systems biology approach for pathway level analysis. Genome Research, 17, 2007.
# Example using a colorectal cancer dataset obtained using Affymetrix geneChip technology (GEE GSE4107). # Suppose that proper preprocessing was performed and a two group moderated t-test was applied. The topTable # result from limma package for this data set is called "top". #The following lines will annotate each probeset to an entrez ID identifier, will keep the most significant probeset for each #gene ID and retain those with FDR<0.05 as differentially expressed. #You can run these lines if hgu133plus2.db package is available #data(colorectalcancer) #x <- hgu133plus2ENTREZID #top$ENTREZ<-unlist(as.list(x[top$ID])) #top<-top[!is.na(top$ENTREZ),] #top<-top[!duplicated(top$ENTREZ),] #tg1<-top[top$adj.P.Val<0.1,] #DE_Colorectal=tg1$logFC #names(DE_Colorectal)<-as.vector(tg1$ENTREZ) #ALL_Colorectal=top$ENTREZ data(colorectalcancer) # pathway analysis using SPIA; # use nB=2000 or higher for more accurate results #uses older version of KEGG signalimng pathways graphs res<-spia(de=DE_Colorectal, all=ALL_Colorectal, organism="hsa",beta=NULL,nB=2000,plots=FALSE, verbose=TRUE,combine="fisher") res # Create the evidence plot plotP(res) #now combine pNDE and pPERT using the normal inversion method without running spia function again res$pG=combfunc(res$pNDE,res$pPERT,combine="norminv") res$pGFdr=p.adjust(res$pG,"fdr") res$pGFWER=p.adjust(res$pG,"bonferroni") plotP(res,threshold=0.05) #highlight the colorectal cancer pathway in green points(I(-log(pPERT))~I(-log(pNDE)),data=res[res$ID=="05210",],col="green",pch=19,cex=1.5) #run SPIA using pathways data generated from (up-to-date) xml files that you can obtain from #KEGG ftp or by downloading them from each pathway's web page: # e.g. go to http://www.genome.jp/kegg/pathway/hsa/hsa04010.html and click on DOwnload KGML #to get the xml file for pathway 4010 makeSPIAdata(kgml.path=system.file("extdata/keggxml/hsa",package="SPIA"),organism="hsa",out.path="./") res<-spia(de=DE_Colorectal, all=ALL_Colorectal, organism="hsa",data.dir="./") res# Example using a colorectal cancer dataset obtained using Affymetrix geneChip technology (GEE GSE4107). # Suppose that proper preprocessing was performed and a two group moderated t-test was applied. The topTable # result from limma package for this data set is called "top". #The following lines will annotate each probeset to an entrez ID identifier, will keep the most significant probeset for each #gene ID and retain those with FDR<0.05 as differentially expressed. #You can run these lines if hgu133plus2.db package is available #data(colorectalcancer) #x <- hgu133plus2ENTREZID #top$ENTREZ<-unlist(as.list(x[top$ID])) #top<-top[!is.na(top$ENTREZ),] #top<-top[!duplicated(top$ENTREZ),] #tg1<-top[top$adj.P.Val<0.1,] #DE_Colorectal=tg1$logFC #names(DE_Colorectal)<-as.vector(tg1$ENTREZ) #ALL_Colorectal=top$ENTREZ data(colorectalcancer) # pathway analysis using SPIA; # use nB=2000 or higher for more accurate results #uses older version of KEGG signalimng pathways graphs res<-spia(de=DE_Colorectal, all=ALL_Colorectal, organism="hsa",beta=NULL,nB=2000,plots=FALSE, verbose=TRUE,combine="fisher") res # Create the evidence plot plotP(res) #now combine pNDE and pPERT using the normal inversion method without running spia function again res$pG=combfunc(res$pNDE,res$pPERT,combine="norminv") res$pGFdr=p.adjust(res$pG,"fdr") res$pGFWER=p.adjust(res$pG,"bonferroni") plotP(res,threshold=0.05) #highlight the colorectal cancer pathway in green points(I(-log(pPERT))~I(-log(pNDE)),data=res[res$ID=="05210",],col="green",pch=19,cex=1.5) #run SPIA using pathways data generated from (up-to-date) xml files that you can obtain from #KEGG ftp or by downloading them from each pathway's web page: # e.g. go to http://www.genome.jp/kegg/pathway/hsa/hsa04010.html and click on DOwnload KGML #to get the xml file for pathway 4010 makeSPIAdata(kgml.path=system.file("extdata/keggxml/hsa",package="SPIA"),organism="hsa",out.path="./") res<-spia(de=DE_Colorectal, all=ALL_Colorectal, organism="hsa",data.dir="./") res
The Vessels dataset consists an named vector DE_Vessels
, which represents the log2 fold changes of the genes chosen as differentially expressed between
ubilical veins and arteries tissue (Kim et al, 2008), and
the universe of all Entrez gene IDs available on the array, ALL_Vessels.
The microarray platform used was Illumina's Human-6 v2 expression BeadChip.
data(Vessels)data(Vessels)
These data was produced at the Perinatology Research Branch, of Wayne State University (Detroit), and
accompanies the publication:
Kim JS, Romero R, Tarca A, Lajeunesse C, Han YM, Kim MJ, Suh YL, Draghici S, Mittal P, Gotsch F, Kusanovic JP, Hassan S, Kim CJ,
Gene expression profiling demonstrates a novel role for fetal fibrocytes and the umbilical vessels in human fetoplacental development,
J Cell Mol Med, 2008, PMID: 18298660.