comparison rgedgeRpaired_nocamera.xml @ 50:60aceade5350 draft

removed atlas - won't compile here at the Baker - thinks we have cpu scaling turned on. We don't afaik.

50:60aceade5350

<tool id="rgDifferentialCount" name="Differential_Count" version="0.23">
  <description>models using BioConductor packages</description>
  <requirements>
      <requirement type="package" version="3.11.11">atlas</requirement>
      <requirement type="package" version="3.0.1">r3</requirement>
      <requirement type="package" version="1.3.18">graphicsmagick</requirement>
      <requirement type="package" version="9.07">ghostscript</requirement>
      <requirement type="package" version="2.12">biocbasics</requirement>
  </requirements>
  
  <command interpreter="python">
     rgToolFactory.py --script_path "$runme" --interpreter "Rscript" --tool_name "DifferentialCounts" 
    --output_dir "$html_file.files_path" --output_html "$html_file" --make_HTML "yes"
  </command>
  <inputs>
    <param name="input1"  type="data" format="tabular" label="Select an input matrix - rows are contigs, columns are counts for each sample"
       help="Use the HTSeq based count matrix preparation tool to create these matrices from BAM/SAM files and a GTF file of genomic features"/>
    <param name="title" type="text" value="Differential Counts" size="80" label="Title for job outputs" 
           help="Supply a meaningful name here to remind you what the outputs contain">
      <sanitizer invalid_char="">
        <valid initial="string.letters,string.digits"><add value="_" /> </valid>
      </sanitizer>
    </param>
    <param name="treatment_name" type="text" value="Treatment" size="50" label="Treatment Name"/>
    <param name="Treat_cols" label="Select columns containing treatment." type="data_column" data_ref="input1" numerical="True" 
         multiple="true" use_header_names="true" size="120" display="checkboxes">
        <validator type="no_options" message="Please select at least one column."/>
        <sanitizer invalid_char="">
          <valid initial="string.letters,string.digits"><add value="_" /> </valid>
        </sanitizer>
    </param>
    <param name="control_name" type="text" value="Control" size="50" label="Control Name"/>
    <param name="Control_cols" label="Select columns containing control." type="data_column" data_ref="input1" numerical="True" 
         multiple="true" use_header_names="true" size="120" display="checkboxes" optional="true">
        <validator type="no_options" message="Please select at least one column."/>
        <sanitizer invalid_char="">
          <valid initial="string.letters,string.digits"><add value="_" /> </valid>
        </sanitizer>        <validator type="no_options" message="Please select at least one column."/>
        <sanitizer invalid_char="">
          <valid initial="string.letters,string.digits"><add value="_" /> </valid>
        </sanitizer>

    </param>
    <param name="subjectids" type="text" optional="true" size="120" value = ""
       label="IF SUBJECTS NOT ALL INDEPENDENT! Enter comma separated strings to indicate sample labels for (eg) pairing - must be one for every column in input"
       help="Leave blank if no pairing, but eg if data from sample id A99 is in columns 2,4 and id C21 is in 3,5 then enter 'A99,C21,A99,C21'">
      <sanitizer>
        <valid initial="string.letters,string.digits"><add value="," /> </valid>
      </sanitizer>
    </param>
    <param name="fQ" type="float" value="0.3" size="5" label="Non-differential contig count quantile threshold - zero to analyze all non-zero read count contigs"
     help="May be a good or a bad idea depending on the biology and the question. EG 0.3 = sparsest 30% of contigs with at least one read are removed before analysis"/>
    <param name="useNDF" type="boolean" truevalue="T" falsevalue="F" checked="false" size="1" 
              label="Non differential filter - remove contigs below a threshold (1 per million) for half or more samples"
     help="May be a good or a bad idea depending on the biology and the question. This was the old default. Quantile based is available as an alternative"/>

    <conditional name="edgeR">
        <param name="doedgeR" type="select" 
           label="Run this model using edgeR"
           help="edgeR uses a negative binomial model and seems to be powerful, even with few replicates">
          <option value="F">Do not run edgeR</option>
          <option value="T" selected="true">Run edgeR</option>
         </param>
         <when value="T">
          <param name="edgeR_priordf" type="integer" value="20" size="3" 
           label="prior.df for tagwise dispersion - lower value = more emphasis on each tag's variance. Replaces prior.n  and prior.df = prior.n * residual.df"
           help="0 = Use edgeR default. Use a small value to 'smooth' small samples. See edgeR docs and note below"/>
         </when>
         <when value="F"></when>
    </conditional>
    <conditional name="DESeq2">
    <param name="doDESeq2" type="select" 
       label="Run the same model with DESeq2 and compare findings"
       help="DESeq2 is an update to the DESeq package. It uses different assumptions and methods to edgeR">
      <option value="F" selected="true">Do not run DESeq2</option>
      <option value="T">Run DESeq2</option>
     </param>
     <when value="T">
         <param name="DESeq_fitType" type="select">
            <option value="parametric" selected="true">Parametric (default) fit for dispersions</option>
            <option value="local">Local fit - this will automagically be used if parametric fit fails</option>
            <option value="mean">Mean dispersion fit- use this if you really understand what you're doing - read the fine manual linked below in the documentation</option>
         </param> 
     </when>
     <when value="F"> </when>
    </conditional>
    <param name="doVoom" type="select" 
       label="Run the same model with Voom/limma and compare findings"
       help="Voom uses counts per million and a precise transformation of variance so count data can be analysed using limma">
      <option value="F" selected="true">Do not run VOOM</option>
      <option value="T">Run VOOM</option>
     </param>
    <!--
    <conditional name="camera">
        <param name="doCamera" type="select" label="Run the edgeR implementation of Camera GSEA for up/down gene sets" 
            help="If yes, you can choose a set of genesets to test and/or supply a gmt format geneset collection from your history">
        <option value="F" selected="true">Do not run GSEA tests with the Camera algorithm</option>
        <option value="T">Run GSEA tests with the Camera algorithm</option>
        </param>
         <when value="T">
         <conditional name="gmtSource">
          <param name="refgmtSource" type="select" 
             label="Use a gene set (.gmt) from your history and/or use a built-in (MSigDB etc) gene set">
            <option value="indexed" selected="true">Use a built-in gene set</option>
            <option value="history">Use a gene set from my history</option>
            <option value="both">Add a gene set from my history to a built in gene set</option>
          </param>
          <when value="indexed">
            <param name="builtinGMT" type="select" label="Select a gene set matrix (.gmt) file to use for the analysis">
              <options from_data_table="gseaGMT_3.1">
                <filter type="sort_by" column="2" />
                <validator type="no_options" message="No GMT v3.1 files are available - please install them"/>
              </options>
            </param>
          </when>
          <when value="history">
            <param name="ownGMT" type="data" format="gmt" label="Select a Gene Set from your history" />
          </when>
          <when value="both">
            <param name="ownGMT" type="data" format="gseagmt" label="Select a Gene Set from your history" />
            <param name="builtinGMT" type="select" label="Select a gene set matrix (.gmt) file to use for the analysis">
              <options from_data_table="gseaGMT_4">
                <filter type="sort_by" column="2" />
                <validator type="no_options" message="No GMT v4 files are available - please fix tool_data_table and loc files"/>
              </options>
            </param>
           </when>
         </conditional>
         </when>
         <when value="F"> 
         </when>
    </conditional>
    -->
    <param name="fdrthresh" type="float" value="0.05" size="5" label="P value threshold for FDR filtering for amily wise error rate control"
     help="Conventional default value of 0.05 recommended"/>
    <param name="fdrtype" type="select" label="FDR (Type II error) control method" 
         help="Use fdr or bh typically to control for the number of tests in a reliable way">
            <option value="fdr" selected="true">fdr</option>
            <option value="BH">Benjamini Hochberg</option>
            <option value="BY">Benjamini Yukateli</option>
            <option value="bonferroni">Bonferroni</option>
            <option value="hochberg">Hochberg</option>
            <option value="holm">Holm</option>
            <option value="hommel">Hommel</option>
            <option value="none">no control for multiple tests</option>
    </param>
  </inputs>
  <outputs>
    <data format="tabular" name="out_edgeR" label="${title}_topTable_edgeR.xls">
         <filter>edgeR['doedgeR'] == "T"</filter>
    </data>
    <data format="tabular" name="out_DESeq2" label="${title}_topTable_DESeq2.xls">
         <filter>DESeq2['doDESeq2'] == "T"</filter>
    </data>
    <data format="tabular" name="out_VOOM" label="${title}_topTable_VOOM.xls">
         <filter>doVoom == "T"</filter>
    </data>
    <data format="html" name="html_file" label="${title}.html"/>
  </outputs>
 <stdio>
     <exit_code range="4"   level="fatal"   description="Number of subject ids must match total number of samples in the input matrix" />
 </stdio>
 <tests>
<test>
<param name='input1' value='test_bams2mx.xls' ftype='tabular' />
 <param name='treatment_name' value='liver' />
 <param name='title' value='edgeRtest' />
 <param name='useNDF' value='' />
 <param name='doedgeR' value='T' />
 <param name='doVoom' value='T' />
 <param name='doDESeq2' value='T' />
 <param name='fdrtype' value='fdr' />
 <param name='edgeR_priordf' value="8" />
 <param name='fdrthresh' value="0.05" />
 <param name='control_name' value='heart' />
 <param name='subjectids' value='' />
 <param name='Control_cols' value='3,4,5,9' />
 <param name='Treat_cols' value='2,6,7,8' />
 <output name='out_edgeR' file='edgeRtest1out.xls' compare='diff' />
 <output name='html_file' file='edgeRtest1out.html'  compare='diff' lines_diff='20' />
</test>
</tests>

<configfiles>
<configfile name="runme">
<![CDATA[
# 
# edgeR.Rscript
# updated npv 2011 for R 2.14.0 and edgeR 2.4.0 by ross
# Performs DGE on a count table containing n replicates of two conditions
#
# Parameters
#
# 1 - Output Dir

# Original edgeR code by: S.Lunke and A.Kaspi
reallybig = log10(.Machine\$double.xmax)
reallysmall = log10(.Machine\$double.xmin)
library('stringr')
library('gplots')
library('edgeR')
hmap2 = function(cmat,nsamp=100,outpdfname='heatmap2.pdf', TName='Treatment',group=NA,myTitle='title goes here')
{
# Perform clustering for significant pvalues after controlling FWER
    samples = colnames(cmat)
    gu = unique(group)
    gn = rownames(cmat)
    if (length(gu) == 2) {
        col.map = function(g) {if (g==gu[1]) "#FF0000" else "#0000FF"}
        pcols = unlist(lapply(group,col.map))        
        } else {
        colours = rainbow(length(gu),start=0,end=4/6)
        pcols = colours[match(group,gu)]        }
    dm = cmat[(! is.na(gn)),] 
    # remove unlabelled hm rows
    nprobes = nrow(dm)
    # sub = paste('Showing',nprobes,'contigs ranked for evidence of differential abundance')
    if (nprobes > nsamp) {
      dm =dm[1:nsamp,]
      #sub = paste('Showing',nsamp,'contigs ranked for evidence for differential abundance out of',nprobes,'total')
    }
    newcolnames = substr(colnames(dm),1,20)
    colnames(dm) = newcolnames
    pdf(outpdfname)
    heatmap.2(dm,main=myTitle,ColSideColors=pcols,col=topo.colors(100),dendrogram="col",key=T,density.info='none',
         Rowv=F,scale='row',trace='none',margins=c(8,8),cexRow=0.4,cexCol=0.5)
    dev.off()
}

hmap = function(cmat,nmeans=4,outpdfname="heatMap.pdf",nsamp=250,TName='Treatment',group=NA,myTitle="Title goes here")
{
    # for 2 groups only was
    #col.map = function(g) {if (g==TName) "#FF0000" else "#0000FF"}
    #pcols = unlist(lapply(group,col.map))
    gu = unique(group)
    colours = rainbow(length(gu),start=0.3,end=0.6)
    pcols = colours[match(group,gu)]
    nrows = nrow(cmat)
    mtitle = paste(myTitle,'Heatmap: n contigs =',nrows)
    if (nrows > nsamp)  {
               cmat = cmat[c(1:nsamp),]
               mtitle = paste('Heatmap: Top ',nsamp,' DE contigs (of ',nrows,')',sep='')
          }
    newcolnames = substr(colnames(cmat),1,20)
    colnames(cmat) = newcolnames
    pdf(outpdfname)
    heatmap(cmat,scale='row',main=mtitle,cexRow=0.3,cexCol=0.4,Rowv=NA,ColSideColors=pcols)
    dev.off()
}

qqPlot = function(descr='qqplot',pvector, outpdf='qqplot.pdf',...)
# stolen from https://gist.github.com/703512
{
    o = -log10(sort(pvector,decreasing=F))
    e = -log10( 1:length(o)/length(o) )
    o[o==-Inf] = reallysmall
    o[o==Inf] = reallybig
    maint = descr
    pdf(outpdf)
    plot(e,o,pch=19,cex=1, main=maint, ...,
        xlab=expression(Expected~~-log[10](italic(p))),
        ylab=expression(Observed~~-log[10](italic(p))),
        xlim=c(0,max(e)), ylim=c(0,max(o)))
    lines(e,e,col="red")
    grid(col = "lightgray", lty = "dotted")
    dev.off()
}

smearPlot = function(DGEList,deTags, outSmear, outMain)
        {
        pdf(outSmear)
        plotSmear(DGEList,de.tags=deTags,main=outMain)
        grid(col="lightgray", lty="dotted")
        dev.off()
        }
        
boxPlot = function(rawrs,cleanrs,maint,myTitle,pdfname)
{   # 
        nc = ncol(rawrs)
        #### for (i in c(1:nc)) {rawrs[(rawrs[,i] < 0),i] = NA}
        fullnames = colnames(rawrs)
        newcolnames = substr(colnames(rawrs),1,20)
        colnames(rawrs) = newcolnames
        newcolnames = substr(colnames(cleanrs),1,20)
        colnames(cleanrs) = newcolnames
        defpar = par(no.readonly=T)
        print.noquote('raw contig counts by sample:')
        print.noquote(summary(rawrs))
        print.noquote('normalised contig counts by sample:')
        print.noquote(summary(cleanrs))
        pdf(pdfname)
        par(mfrow=c(1,2))
        boxplot(rawrs,varwidth=T,notch=T,ylab='log contig count',col="maroon",las=3,cex.axis=0.35,main=paste('Raw:',maint))
        grid(col="lightgray",lty="dotted")
        boxplot(cleanrs,varwidth=T,notch=T,ylab='log contig count',col="maroon",las=3,cex.axis=0.35,main=paste('After ',maint))
        grid(col="lightgray",lty="dotted")
        dev.off()
        pdfname = "sample_counts_histogram.pdf" 
        nc = ncol(rawrs)
        print.noquote(paste('Using ncol rawrs=',nc))
        ncroot = round(sqrt(nc))
        if (ncroot*ncroot < nc) { ncroot = ncroot + 1 }
        m = c()
        for (i in c(1:nc)) {
              rhist = hist(rawrs[,i],breaks=100,plot=F)
              m = append(m,max(rhist\$counts))
             }
        ymax = max(m)
        ncols = length(fullnames)
        if (ncols > 20) 
        {
           scale = 7*ncols/20
           pdf(pdfname,width=scale,height=scale)
        } else { 
           pdf(pdfname)
        }
        par(mfrow=c(ncroot,ncroot))
        for (i in c(1:nc)) {
                 hist(rawrs[,i], main=paste("Contig logcount",i), xlab='log raw count', col="maroon", 
                 breaks=100,sub=fullnames[i],cex=0.8,ylim=c(0,ymax))
             }
        dev.off()
        par(defpar)
      
}

cumPlot = function(rawrs,cleanrs,maint,myTitle)
{   # updated to use ecdf
        pdfname = "Filtering_rowsum_bar_charts.pdf"
        defpar = par(no.readonly=T)
        lrs = log(rawrs,10) 
        lim = max(lrs)
        pdf(pdfname)
        par(mfrow=c(2,1))
        hist(lrs,breaks=100,main=paste('Before:',maint),xlab="# Reads (log)",
             ylab="Count",col="maroon",sub=myTitle, xlim=c(0,lim),las=1)
        grid(col="lightgray", lty="dotted")
        lrs = log(cleanrs,10) 
        hist(lrs,breaks=100,main=paste('After:',maint),xlab="# Reads (log)",
             ylab="Count",col="maroon",sub=myTitle,xlim=c(0,lim),las=1)
        grid(col="lightgray", lty="dotted")
        dev.off()
        par(defpar)
}

cumPlot1 = function(rawrs,cleanrs,maint,myTitle)
{   # updated to use ecdf
        pdfname = paste(gsub(" ","", myTitle , fixed=TRUE),"RowsumCum.pdf",sep='_')
        pdf(pdfname)
        par(mfrow=c(2,1))
        lastx = max(rawrs)
        rawe = knots(ecdf(rawrs))
        cleane = knots(ecdf(cleanrs))
        cy = 1:length(cleane)/length(cleane)
        ry = 1:length(rawe)/length(rawe)
        plot(rawe,ry,type='l',main=paste('Before',maint),xlab="Log Contig Total Reads",
             ylab="Cumulative proportion",col="maroon",log='x',xlim=c(1,lastx),sub=myTitle)
        grid(col="blue")
        plot(cleane,cy,type='l',main=paste('After',maint),xlab="Log Contig Total Reads",
             ylab="Cumulative proportion",col="maroon",log='x',xlim=c(1,lastx),sub=myTitle)
        grid(col="blue")
        dev.off()
}



doGSEA = function(y=NULL,design=NULL,histgmt="",
                  bigmt="/data/genomes/gsea/3.1/Abetterchoice_nocgp_c2_c3_c5_symbols_all.gmt",
                  ntest=0, myTitle="myTitle", outfname="GSEA.xls", minnin=5, maxnin=2000,fdrthresh=0.05,fdrtype="BH")
{
  sink('Camera.log')
  genesets = c()
  if (bigmt > "")
  {
    bigenesets = readLines(bigmt)
    genesets = bigenesets
  }
  if (histgmt > "")
  {
    hgenesets = readLines(histgmt)
    if (bigmt > "") {
      genesets = rbind(genesets,hgenesets)
    } else {
      genesets = hgenesets
    } # use only history if no bi
  }
  print.noquote(paste("@@@read",length(genesets), 'genesets from',histgmt,bigmt))
  genesets = strsplit(genesets,'\t') # tabular. genesetid\tURLorwhatever\tgene_1\t..\tgene_n
  outf = outfname
  head=paste(myTitle,'edgeR GSEA')
  write(head,file=outfname,append=F)
  ntest=length(genesets)
  urownames = toupper(rownames(y))
  upcam = c()
  downcam = c()
  for (i in 1:ntest) {
    gs = unlist(genesets[i])
    g = gs[1] # geneset_id
    u = gs[2]
    if (u > "") { u = paste("<a href=\'",u,"\'>",u,"</a>",sep="") }
    glist = gs[3:length(gs)] # member gene symbols
    glist = toupper(glist)
    inglist = urownames %in% glist
    nin = sum(inglist)
    if ((nin > minnin) && (nin < maxnin)) {
      ### print(paste('@@found',sum(inglist),'genes in glist'))
      camres = camera(y=y,index=inglist,design=design)
      if (! is.null(camres)) {
      rownames(camres) = g # gene set name
      camres = cbind(GeneSet=g,URL=u,camres)
      if (camres\$Direction == "Up") 
        {
        upcam = rbind(upcam,camres) } else {
          downcam = rbind(downcam,camres)
        }
      }
   }
  }
  uscam = upcam[order(upcam\$PValue),]
  unadjp = uscam\$PValue
  uscam\$adjPValue = p.adjust(unadjp,method=fdrtype)
  nup = max(10,sum((uscam\$adjPValue < fdrthresh)))
  dscam = downcam[order(downcam\$PValue),]
  unadjp = dscam\$PValue
  dscam\$adjPValue = p.adjust(unadjp,method=fdrtype)
  ndown = max(10,sum((dscam\$adjPValue < fdrthresh)))
  write.table(uscam,file=paste('camera_up',outfname,sep='_'),quote=F,sep='\t',row.names=F)
  write.table(dscam,file=paste('camera_down',outfname,sep='_'),quote=F,sep='\t',row.names=F)
  print.noquote(paste('@@@@@ Camera up top',nup,'gene sets:'))
  write.table(head(uscam,nup),file="",quote=F,sep='\t',row.names=F)
  print.noquote(paste('@@@@@ Camera down top',ndown,'gene sets:'))
  write.table(head(dscam,ndown),file="",quote=F,sep='\t',row.names=F)
  sink()
}




doGSEAatonce = function(y=NULL,design=NULL,histgmt="",
                  bigmt="/data/genomes/gsea/3.1/Abetterchoice_nocgp_c2_c3_c5_symbols_all.gmt",
                  ntest=0, myTitle="myTitle", outfname="GSEA.xls", minnin=5, maxnin=2000,fdrthresh=0.05,fdrtype="BH")
{
  sink('Camera.log')
  genesets = c()
  if (bigmt > "")
  {
    bigenesets = readLines(bigmt)
    genesets = bigenesets
  }
  if (histgmt > "")
  {
    hgenesets = readLines(histgmt)
    if (bigmt > "") {
      genesets = rbind(genesets,hgenesets)
    } else {
      genesets = hgenesets
    } # use only history if no bi
  }
  print.noquote(paste("@@@read",length(genesets), 'genesets from',histgmt,bigmt))
  genesets = strsplit(genesets,'\t') # tabular. genesetid\tURLorwhatever\tgene_1\t..\tgene_n
  outf = outfname
  head=paste(myTitle,'edgeR GSEA')
  write(head,file=outfname,append=F)
  ntest=length(genesets)
  urownames = toupper(rownames(y))
  upcam = c()
  downcam = c()
  incam = c()
  urls = c()
  gsids = c()
  for (i in 1:ntest) {
    gs = unlist(genesets[i])
    gsid = gs[1] # geneset_id
    url = gs[2]
    if (url > "") { url = paste("<a href=\'",url,"\'>",url,"</a>",sep="") }
    glist = gs[3:length(gs)] # member gene symbols
    glist = toupper(glist)
    inglist = urownames %in% glist
    nin = sum(inglist)
    if ((nin > minnin) && (nin < maxnin)) {
      incam = c(incam,inglist)
      gsids = c(gsids,gsid)
      urls = c(urls,url)
    }
  }
  incam = as.list(incam)
  names(incam) = gsids
  allcam = camera(y=y,index=incam,design=design)
  allcamres = cbind(geneset=gsids,allcam,URL=urls)
  for (i in 1:ntest) {
    camres = allcamres[i]
    res = try(test = (camres\$Direction == "Up"))
    if ("try-error" %in% class(res)) {
      cat("test failed, camres = :")
      print.noquote(camres)
    } else  { if (camres\$Direction == "Up")
    {  upcam = rbind(upcam,camres)
    } else { downcam = rbind(downcam,camres)
    }
              
    }
  }
  uscam = upcam[order(upcam\$PValue),]
  unadjp = uscam\$PValue
  uscam\$adjPValue = p.adjust(unadjp,method=fdrtype)
  nup = max(10,sum((uscam\$adjPValue < fdrthresh)))
  dscam = downcam[order(downcam\$PValue),]
  unadjp = dscam\$PValue
  dscam\$adjPValue = p.adjust(unadjp,method=fdrtype)
  ndown = max(10,sum((dscam\$adjPValue < fdrthresh)))
  write.table(uscam,file=paste('camera_up',outfname,sep='_'),quote=F,sep='\t',row.names=F)
  write.table(dscam,file=paste('camera_down',outfname,sep='_'),quote=F,sep='\t',row.names=F)
  print.noquote(paste('@@@@@ Camera up top',nup,'gene sets:'))
  write.table(head(uscam,nup),file="",quote=F,sep='\t',row.names=F)
  print.noquote(paste('@@@@@ Camera down top',ndown,'gene sets:'))
  write.table(head(dscam,ndown),file="",quote=F,sep='\t',row.names=F)
  sink()
  }
 

edgeIt = function (Count_Matrix=c(),group=c(),out_edgeR=F,out_VOOM=F,out_DESeq2=F,fdrtype='fdr',priordf=5, 
        fdrthresh=0.05,outputdir='.', myTitle='Differential Counts',libSize=c(),useNDF=F,
        filterquantile=0.2, subjects=c(),mydesign=NULL,
        doDESeq2=T,doVoom=T,doCamera=T,doedgeR=T,org='hg19',
        histgmt="", bigmt="/data/genomes/gsea/3.1/Abetterchoice_nocgp_c2_c3_c5_symbols_all.gmt",
        doCook=F,DESeq_fitType="parameteric") 
{
  # Error handling
  if (length(unique(group))!=2){
    print("Number of conditions identified in experiment does not equal 2")
    q()
    }
  require(edgeR)
  options(width = 512) 
  mt = paste(unlist(strsplit(myTitle,'_')),collapse=" ")
  allN = nrow(Count_Matrix)
  nscut = round(ncol(Count_Matrix)/2)
  colTotmillionreads = colSums(Count_Matrix)/1e6
  counts.dataframe = as.data.frame(c()) 
  rawrs = rowSums(Count_Matrix)
  nonzerod = Count_Matrix[(rawrs > 0),] # remove all zero count genes
  nzN = nrow(nonzerod)
  nzrs = rowSums(nonzerod)
  zN = allN - nzN
  print('# Quantiles for non-zero row counts:',quote=F)
  print(quantile(nzrs,probs=seq(0,1,0.1)),quote=F)
  if (useNDF == T)
  {
    gt1rpin3 = rowSums(Count_Matrix/expandAsMatrix(colTotmillionreads,dim(Count_Matrix)) >= 1) >= nscut
    lo = colSums(Count_Matrix[!gt1rpin3,])
    workCM = Count_Matrix[gt1rpin3,]
    cleanrs = rowSums(workCM)
    cleanN = length(cleanrs)
    meth = paste( "After removing",length(lo),"contigs with fewer than ",nscut," sample read counts >= 1 per million, there are",sep="")
    print(paste("Read",allN,"contigs. Removed",zN,"contigs with no reads.",meth,cleanN,"contigs"),quote=F)
    maint = paste('Filter >=1/million reads in >=',nscut,'samples')
  }   else {        
    useme = (nzrs > quantile(nzrs,filterquantile))
    workCM = nonzerod[useme,]
    lo = colSums(nonzerod[!useme,])
    cleanrs = rowSums(workCM)
    cleanN = length(cleanrs)
    meth = paste("After filtering at count quantile =",filterquantile,", there are",sep="")
    print(paste('Read',allN,"contigs. Removed",zN,"with no reads.",meth,cleanN,"contigs"),quote=F)
    maint = paste('Filter below',filterquantile,'quantile')
  }
  cumPlot(rawrs=rawrs,cleanrs=cleanrs,maint=maint,myTitle=myTitle)
  allgenes = rownames(workCM)
  reg = "^chr([0-9]+):([0-9]+)-([0-9]+)"
  genecards="<a href=\'http://www.genecards.org/index.php?path=/Search/keyword/"
  ucsc = paste("<a href=\'http://genome.ucsc.edu/cgi-bin/hgTracks?db=",org,sep='')
  testreg = str_match(allgenes,reg)
  if (sum(!is.na(testreg[,1]))/length(testreg[,1]) > 0.8) # is ucsc style string
  {
    print("@@ using ucsc substitution for urls")
    contigurls = paste0(ucsc,"&amp;position=chr",testreg[,2],":",testreg[,3],"-",testreg[,4],"\'>",allgenes,"</a>")
  } else {
    print.noquote("@@ using genecards substitution for urls")
    contigurls = paste0(genecards,allgenes,"\'>",allgenes,"</a>")
  }
  print(paste("# Total low count contigs per sample = ",paste(lo,collapse=',')),quote=F) 
  cmrowsums = rowSums(workCM)
  TName=unique(group)[1]
  CName=unique(group)[2]
  if (is.null(mydesign)) {
    if (length(subjects) == 0) 
    {
      mydesign = model.matrix(~group)
    } 
    else { 
      subjf = factor(subjects)
      mydesign = model.matrix(~subjf+group) # we block on subject so make group last to simplify finding it
    }
  } 
  print.noquote(paste('Using samples:',paste(colnames(workCM),collapse=',')))
  print.noquote('Using design matrix:')
  print.noquote(mydesign)
  if (doedgeR) {
  sink('edgeR.log')
  #### Setup DGEList object
  DGEList = DGEList(counts=workCM, group = group)
  DGEList = calcNormFactors(DGEList)

  DGEList = estimateGLMCommonDisp(DGEList,mydesign)
  comdisp = DGEList\$common.dispersion
  DGEList = estimateGLMTrendedDisp(DGEList,mydesign)
  if (edgeR_priordf > 0) {
    print.noquote(paste("prior.df =",edgeR_priordf))
    DGEList = estimateGLMTagwiseDisp(DGEList,mydesign,prior.df = edgeR_priordf)
  } else {
    DGEList = estimateGLMTagwiseDisp(DGEList,mydesign)
  }
  DGLM = glmFit(DGEList,design=mydesign)
  DE = glmLRT(DGLM,coef=ncol(DGLM\$design)) # always last one - subject is first if needed
  efflib = DGEList\$samples\$lib.size*DGEList\$samples\$norm.factors
  normData = (1e+06*DGEList\$counts/efflib)
  uoutput = cbind( 
    Name=as.character(rownames(DGEList\$counts)),
    DE\$table,
    adj.p.value=p.adjust(DE\$table\$PValue, method=fdrtype),
    Dispersion=DGEList\$tagwise.dispersion,totreads=cmrowsums,normData,
    DGEList\$counts
  )
  soutput = uoutput[order(DE\$table\$PValue),] # sorted into p value order - for quick toptable
  goodness = gof(DGLM, pcutoff=fdrthresh)
  if (sum(goodness\$outlier) > 0) {
    print.noquote('GLM outliers:')
    print(paste(rownames(DGLM)[(goodness\$outlier)],collapse=','),quote=F)
    } else { 
      print('No GLM fit outlier genes found\n')
    }
  z = limma::zscoreGamma(goodness\$gof.statistic, shape=goodness\$df/2, scale=2)
  pdf("edgeR_GoodnessofFit.pdf")
  qq = qqnorm(z, panel.first=grid(), main="tagwise dispersion")
  abline(0,1,lwd=3)
  points(qq\$x[goodness\$outlier],qq\$y[goodness\$outlier], pch=16, col="maroon")
  dev.off()
  estpriorn = getPriorN(DGEList)
  print(paste("Common Dispersion =",comdisp,"CV = ",sqrt(comdisp),"getPriorN = ",estpriorn),quote=F)
  efflib = DGEList\$samples\$lib.size*DGEList\$samples\$norm.factors
  normData = (1e+06*DGEList\$counts)/efflib
  lnormData = log(normData + 1e-6,10)
  uniqueg = unique(group)
  #### Plot MDS
  sample_colors =  match(group,levels(group))
  sampleTypes = levels(factor(group))
  print.noquote(sampleTypes)
  pdf("edgeR_MDSplot.pdf")
  plotMDS.DGEList(DGEList,main=paste("edgeR MDS for",myTitle),cex=0.5,col=sample_colors,pch=sample_colors)
  legend(x="topleft", legend = sampleTypes,col=c(1:length(sampleTypes)), pch=19)
  grid(col="blue")
  dev.off()
  colnames(normData) = paste( colnames(normData),'N',sep="_")
  print(paste('Raw sample read totals',paste(colSums(nonzerod,na.rm=T),collapse=',')))
  nzd = data.frame(log(nonzerod + 1e-2,10))
  try( boxPlot(rawrs=nzd,cleanrs=lnormData,maint='TMM Normalisation',myTitle=myTitle,pdfname="edgeR_raw_norm_counts_box.pdf") )
  write.table(soutput,file=out_edgeR, quote=FALSE, sep="\t",row.names=F)
  tt = cbind( 
    Name=as.character(rownames(DGEList\$counts)),
    DE\$table,
    adj.p.value=p.adjust(DE\$table\$PValue, method=fdrtype),
    Dispersion=DGEList\$tagwise.dispersion,totreads=cmrowsums
  )
  print.noquote("# edgeR Top tags\n")
  tt = cbind(tt,URL=contigurls) # add to end so table isn't laid out strangely
  tt = tt[order(DE\$table\$PValue),] 
  print.noquote(tt[1:50,])
  deTags = rownames(uoutput[uoutput\$adj.p.value < fdrthresh,])
  nsig = length(deTags)
  print(paste('#',nsig,'tags significant at adj p=',fdrthresh),quote=F)
  deColours = ifelse(deTags,'red','black')
  pdf("edgeR_BCV_vs_abundance.pdf")
  plotBCV(DGEList, cex=0.3, main="Biological CV vs abundance",col.tagwise=deColours)
  dev.off()
  dg = DGEList[order(DE\$table\$PValue),]
  #normData = (1e+06 * dg\$counts/expandAsMatrix(dg\$samples\$lib.size, dim(dg)))
  efflib = dg\$samples\$lib.size*dg\$samples\$norm.factors
  normData = (1e+06*dg\$counts/efflib)
  outpdfname="edgeR_top_100_heatmap.pdf"
  hmap2(normData,nsamp=100,TName=TName,group=group,outpdfname=outpdfname,myTitle=paste('edgeR Heatmap',myTitle))
  outSmear = "edgeR_smearplot.pdf"
  outMain = paste("Smear Plot for ",TName,' Vs ',CName,' (FDR@',fdrthresh,' N = ',nsig,')',sep='')
  smearPlot(DGEList=DGEList,deTags=deTags, outSmear=outSmear, outMain = outMain)
  qqPlot(descr=paste(myTitle,'edgeR adj p QQ plot'),pvector=tt\$adj.p.value,outpdf='edgeR_qqplot.pdf')
  norm.factor = DGEList\$samples\$norm.factors
  topresults.edgeR = soutput[which(soutput\$adj.p.value < fdrthresh), ]
  edgeRcountsindex = which(allgenes %in% rownames(topresults.edgeR))
  edgeRcounts = rep(0, length(allgenes))
  edgeRcounts[edgeRcountsindex] = 1  # Create venn diagram of hits
  sink()
  } ### doedgeR
  if (doDESeq2 == T)
  {
    sink("DESeq2.log")
    # DESeq2
    require('DESeq2')
    library('RColorBrewer')
    if (length(subjects) == 0)
        {
        pdata = data.frame(Name=colnames(workCM),Rx=group,row.names=colnames(workCM))
        deSEQds = DESeqDataSetFromMatrix(countData = workCM,  colData = pdata, design = formula(~ Rx))
        } else {
        pdata = data.frame(Name=colnames(workCM),Rx=group,subjects=subjects,row.names=colnames(workCM))
        deSEQds = DESeqDataSetFromMatrix(countData = workCM,  colData = pdata, design = formula(~ subjects + Rx))
        }
    #DESeq2 = DESeq(deSEQds,fitType='local',pAdjustMethod=fdrtype)
    #rDESeq = results(DESeq2)
    #newCountDataSet(workCM, group)
    deSeqDatsizefac = estimateSizeFactors(deSEQds)
    deSeqDatdisp = estimateDispersions(deSeqDatsizefac,fitType=DESeq_fitType)
    resDESeq = nbinomWaldTest(deSeqDatdisp, pAdjustMethod=fdrtype)
    rDESeq = as.data.frame(results(resDESeq))
    rDESeq = cbind(Contig=rownames(workCM),rDESeq,NReads=cmrowsums)
    srDESeq = rDESeq[order(rDESeq\$pvalue),]
    write.table(srDESeq,file=out_DESeq2, quote=FALSE, sep="\t",row.names=F)
    qqPlot(descr=paste(myTitle,'DESeq2 adj p qq plot'),pvector=rDESeq\$padj,outpdf='DESeq2_qqplot.pdf')
    cat("# DESeq top 50\n")
    rDESeq = cbind(Contig=rownames(workCM),rDESeq,NReads=cmrowsums,URL=contigurls)
    srDESeq = rDESeq[order(rDESeq\$pvalue),]
    print.noquote(srDESeq[1:50,])
    topresults.DESeq = rDESeq[which(rDESeq\$padj < fdrthresh), ]
    DESeqcountsindex = which(allgenes %in% rownames(topresults.DESeq))
    DESeqcounts = rep(0, length(allgenes))
    DESeqcounts[DESeqcountsindex] = 1
    pdf("DESeq2_dispersion_estimates.pdf")
    plotDispEsts(resDESeq)
    dev.off()
    ysmall = abs(min(rDESeq\$log2FoldChange))
    ybig = abs(max(rDESeq\$log2FoldChange))
    ylimit = min(4,ysmall,ybig)
    pdf("DESeq2_MA_plot.pdf")
    plotMA(resDESeq,main=paste(myTitle,"DESeq2 MA plot"),ylim=c(-ylimit,ylimit))
    dev.off()
    rlogres = rlogTransformation(resDESeq)
    sampledists = dist( t( assay(rlogres) ) )
    sdmat = as.matrix(sampledists)
    pdf("DESeq2_sample_distance_plot.pdf")
    heatmap.2(sdmat,trace="none",main=paste(myTitle,"DESeq2 sample distances"),
         col = colorRampPalette( rev(brewer.pal(9, "RdBu")) )(255))
    dev.off()
    ###outpdfname="DESeq2_top50_heatmap.pdf"
    ###hmap2(sresDESeq,nsamp=50,TName=TName,group=group,outpdfname=outpdfname,myTitle=paste('DESeq2 vst rlog Heatmap',myTitle))
    sink()
    result = try( (ppca = plotPCA( varianceStabilizingTransformation(deSeqDatdisp,blind=T), intgroup=c("Rx","Name")) ) )
    if ("try-error" %in% class(result)) {
         print.noquote('DESeq2 plotPCA failed.')
         } else {
         pdf("DESeq2_PCA_plot.pdf")
         #### wtf - print? Seems needed to get this to work
         print(ppca)
         dev.off()
        }
  }

  if (doVoom == T) {
      sink('Voom.log')
      if (doedgeR == F) {
         #### Setup DGEList object
         DGEList = DGEList(counts=workCM, group = group)
         DGEList = calcNormFactors(DGEList)
         DGEList = estimateGLMCommonDisp(DGEList,mydesign)
         DGEList = estimateGLMTrendedDisp(DGEList,mydesign)
         DGEList = estimateGLMTagwiseDisp(DGEList,mydesign)
         DGEList = estimateGLMTagwiseDisp(DGEList,mydesign)
         norm.factor = DGEList\$samples\$norm.factors
         }
      pdf("Voom_mean_variance_plot.pdf")
      dat.voomed = voom(DGEList, mydesign, plot = TRUE, lib.size = colSums(workCM) * norm.factor)
      dev.off()
      # Use limma to fit data
      fit = lmFit(dat.voomed, mydesign)
      fit = eBayes(fit)
      rvoom = topTable(fit, coef = length(colnames(mydesign)), adj = fdrtype, n = Inf, sort="none")
      qqPlot(descr=paste(myTitle,'Voom-limma adj p QQ plot'),pvector=rvoom\$adj.P.Val,outpdf='Voom_qqplot.pdf')
      rownames(rvoom) = rownames(workCM)
      rvoom = cbind(rvoom,NReads=cmrowsums)
      srvoom = rvoom[order(rvoom\$P.Value),]
      write.table(srvoom,file=out_VOOM, quote=FALSE, sep="\t",row.names=F)
      rvoom = cbind(rvoom,URL=contigurls)
      deTags = rownames(rvoom[rvoom\$adj.p.value < fdrthresh,])    
      nsig = length(deTags)  
      cat("# Voom top 50\n")
      print(srvoom[1:50,])
	  normData = srvoom\$E
	  outpdfname="Voom_top_100_heatmap.pdf"
	  hmap2(normData,nsamp=100,TName=TName,group=group,outpdfname=outpdfname,myTitle=paste('VOOM Heatmap',myTitle))
	  outSmear = "Voom_smearplot.pdf"
	  outMain = paste("Smear Plot for ",TName,' Vs ',CName,' (FDR@',fdrthresh,' N = ',nsig,')',sep='')
	  smearPlot(DGEList=rvoom,deTags=deTags, outSmear=outSmear, outMain = outMain)
	  qqPlot(descr=paste(myTitle,'VOOM adj p QQ plot'),pvector=srvoom\$adj.P.Val,outpdf='Voom_qqplot.pdf')
      # Use an FDR cutoff to find interesting samples for edgeR, DESeq and voom/limma
      topresults.voom = rvoom[which(rvoom\$adj.P.Val < fdrthresh), ]
      voomcountsindex = which(allgenes %in% topresults.voom\$ID)
      voomcounts = rep(0, length(allgenes))
      voomcounts[voomcountsindex] = 1
      sink()
  }

  if (doCamera) {
  doGSEA(y=DGEList,design=mydesign,histgmt=histgmt,bigmt=bigmt,ntest=20,myTitle=myTitle,
    outfname=paste(mt,"GSEA.xls",sep="_"),fdrthresh=fdrthresh,fdrtype=fdrtype)
  }

  if ((doDESeq2==T) || (doVoom==T) || (doedgeR==T)) {
    if ((doVoom==T) && (doDESeq2==T) && (doedgeR==T)) {
        vennmain = paste(mt,'Voom,edgeR and DESeq2 overlap at FDR=',fdrthresh)
        counts.dataframe = data.frame(edgeR = edgeRcounts, DESeq2 = DESeqcounts, 
                                       VOOM_limma = voomcounts, row.names = allgenes)
       } else if ((doDESeq2==T) && (doedgeR==T))  {
         vennmain = paste(mt,'DESeq2 and edgeR overlap at FDR=',fdrthresh)
         counts.dataframe = data.frame(edgeR = edgeRcounts, DESeq2 = DESeqcounts, row.names = allgenes)
       } else if ((doVoom==T) && (doedgeR==T)) {
        vennmain = paste(mt,'Voom and edgeR overlap at FDR=',fdrthresh)
        counts.dataframe = data.frame(edgeR = edgeRcounts, VOOM_limma = voomcounts, row.names = allgenes)
       }
    
    if (nrow(counts.dataframe > 1)) {
      counts.venn = vennCounts(counts.dataframe)
      vennf = "Venn_significant_genes_overlap.pdf" 
      pdf(vennf)
      vennDiagram(counts.venn,main=vennmain,col="maroon")
      dev.off()
    }
  } #### doDESeq2 or doVoom
  
}
#### Done

###sink(stdout(),append=T,type="message")
builtin_gmt = ""
history_gmt = ""
history_gmt_name = ""
out_edgeR = F
out_DESeq2 = F
out_VOOM = "$out_VOOM"
doDESeq2 = $DESeq2.doDESeq2 # make these T or F
doVoom = $doVoom
doCamera = F
doedgeR = $edgeR.doedgeR
edgeR_priordf = 0


#if $doVoom == "T":
  out_VOOM = "$out_VOOM"
#end if

#if $DESeq2.doDESeq2 == "T":
  out_DESeq2 = "$out_DESeq2"
  DESeq_fitType = "$DESeq2.DESeq_fitType"
#end if

#if $edgeR.doedgeR == "T":
  out_edgeR = "$out_edgeR"
  edgeR_priordf = $edgeR.edgeR_priordf  
#end if


if (sum(c(doedgeR,doVoom,doDESeq2)) == 0)
{
write("No methods chosen - nothing to do! Please try again after choosing one or more methods", stderr())
quit(save="no",status=2)
}

Out_Dir = "$html_file.files_path"
Input =  "$input1"
TreatmentName = "$treatment_name"
TreatmentCols = "$Treat_cols"
ControlName = "$control_name"
ControlCols= "$Control_cols"
org = "$input1.dbkey"
if (org == "") { org = "hg19"}
fdrtype = "$fdrtype"
fdrthresh = $fdrthresh
useNDF = $useNDF
fQ = $fQ # non-differential centile cutoff
myTitle = "$title"
sids = strsplit("$subjectids",',')
subjects = unlist(sids)
nsubj = length(subjects)
TCols = as.numeric(strsplit(TreatmentCols,",")[[1]])-1
CCols = as.numeric(strsplit(ControlCols,",")[[1]])-1 
cat('Got TCols=')
cat(TCols)
cat('; CCols=')
cat(CCols)
cat('\n')
useCols = c(TCols,CCols)
if (file.exists(Out_Dir) == F) dir.create(Out_Dir)
Count_Matrix = read.table(Input,header=T,row.names=1,sep='\t') #Load tab file assume header
snames = colnames(Count_Matrix)
nsamples = length(snames)
if (nsubj >  0 & nsubj != nsamples) {
options("show.error.messages"=T)
mess = paste('Fatal error: Supplied subject id list',paste(subjects,collapse=','),
   'has length',nsubj,'but there are',nsamples,'samples',paste(snames,collapse=','))
write(mess, stderr())
quit(save="no",status=4)
}
if (length(subjects) != 0) {subjects = subjects[useCols]}
Count_Matrix = Count_Matrix[,useCols] ### reorder columns
rn = rownames(Count_Matrix)
islib = rn %in% c('librarySize','NotInBedRegions')
LibSizes = Count_Matrix[subset(rn,islib),][1] # take first
Count_Matrix = Count_Matrix[subset(rn,! islib),]
group = c(rep(TreatmentName,length(TCols)), rep(ControlName,length(CCols)) )             #Build a group descriptor
group = factor(group, levels=c(ControlName,TreatmentName))
colnames(Count_Matrix) = paste(group,colnames(Count_Matrix),sep="_")                   #Relable columns
results = edgeIt(Count_Matrix=Count_Matrix,group=group, out_edgeR=out_edgeR, out_VOOM=out_VOOM, out_DESeq2=out_DESeq2,
                 fdrtype='BH',mydesign=NULL,priordf=edgeR_priordf,fdrthresh=fdrthresh,outputdir='.',
                 myTitle=myTitle,useNDF=F,libSize=c(),filterquantile=fQ,subjects=subjects,
                 doDESeq2=doDESeq2,doVoom=doVoom,doCamera=doCamera,doedgeR=doedgeR,org=org,
                 histgmt=history_gmt,bigmt=builtin_gmt,DESeq_fitType=DESeq_fitType)
sessionInfo()
]]>
</configfile>
</configfiles>
<help>

**What it does**

Allows short read sequence counts from controlled experiments to be analysed for differentially expressed genes.
Optionally adds a term for subject if not all samples are independent or if some other factor needs to be blocked in the design.

**Input**

Requires a count matrix as a tabular file. These are best made using the companion HTSeq_ based counter Galaxy wrapper
and your fave gene model to generate inputs. Each row is a genomic feature (gene or exon eg) and each column the 
non-negative integer count of reads from one sample overlapping the feature. 
The matrix must have a header row uniquely identifying the source samples, and unique row names in 
the first column. Typically the row names are gene symbols or probe ids for downstream use in GSEA and other methods.

**Specifying comparisons**

This is basically dumbed down for two factors - case vs control.

More complex interfaces are possible but painful at present. 
Probably need to specify a phenotype file to do this better.
Work in progress. Send code.

If you have (eg) paired samples and wish to include a term in the GLM to account for some other factor (subject in the case of paired samples),
put a comma separated list of indicators for every sample (whether modelled or not!) indicating (eg) the subject number or 
A list of integers, one for each subject or an empty string if samples are all independent.
If not empty, there must be exactly as many integers in the supplied integer list as there are columns (samples) in the count matrix.
Integers for samples that are not in the analysis *must* be present in the string as filler even if not used.

So if you have 2 pairs out of 6 samples, you need to put in unique integers for the unpaired ones
eg if you had 6 samples with the first two independent but the second and third pairs each being from independent subjects. you might use
8,9,1,1,2,2 
as subject IDs to indicate two paired samples from the same subject in columns 3/4 and 5/6

**Methods available**

You can run 3 popular Bioconductor packages available for count data.

edgeR - see edgeR_ for details

VOOM/limma - see limma_VOOM_ for details

DESeq2 - see DESeq2_ for details

and optionally camera in edgeR which works better if MSigDB is installed.

**Outputs**

Some helpful plots and analysis results. Note that most of these are produced using R code 
suggested by the excellent documentation and vignettes for the Bioconductor
packages invoked. The Tool Factory is used to automatically lay these out for you to enjoy.

**Note on Voom**

The voom from limma version 3.16.6 help in R includes this from the authors - but you should read the paper to interpret this method.

This function is intended to process RNA-Seq or ChIP-Seq data prior to linear modelling in limma.

voom is an acronym for mean-variance modelling at the observational level.
The key concern is to estimate the mean-variance relationship in the data, then use this to compute appropriate weights for each observation.
Count data almost show non-trivial mean-variance relationships. Raw counts show increasing variance with increasing count size, while log-counts typically show a decreasing mean-variance trend.
This function estimates the mean-variance trend for log-counts, then assigns a weight to each observation based on its predicted variance.
The weights are then used in the linear modelling process to adjust for heteroscedasticity.

In an experiment, a count value is observed for each tag in each sample. A tag-wise mean-variance trend is computed using lowess.
The tag-wise mean is the mean log2 count with an offset of 0.5, across samples for a given tag.
The tag-wise variance is the quarter-root-variance of normalized log2 counts per million values with an offset of 0.5, across samples for a given tag.
Tags with zero counts across all samples are not included in the lowess fit. Optional normalization is performed using normalizeBetweenArrays.
Using fitted values of log2 counts from a linear model fit by lmFit, variances from the mean-variance trend were interpolated for each observation.
This was carried out by approxfun. Inverse variance weights can be used to correct for mean-variance trend in the count data.


Author(s)

Charity Law and Gordon Smyth

References

Law, CW (2013). Precision weights for gene expression analysis. PhD Thesis. University of Melbourne, Australia.

Law, CW, Chen, Y, Shi, W, Smyth, GK (2013). Voom! Precision weights unlock linear model analysis tools for RNA-seq read counts.
Technical Report 1 May 2013, Bioinformatics Division, Walter and Eliza Hall Institute of Medical Reseach, Melbourne, Australia.
http://www.statsci.org/smyth/pubs/VoomPreprint.pdf

See Also

A voom case study is given in the edgeR User's Guide.

vooma is a similar function but for microarrays instead of RNA-seq.


***old rant on changes to Bioconductor package variable names between versions***

The edgeR authors made a small cosmetic change in the name of one important variable (from p.value to PValue) 
breaking this and all other code that assumed the old name for this variable, 
between edgeR2.4.4 and 2.4.6 (the version for R 2.14 as at the time of writing). 
This means that all code using edgeR is sensitive to the version. I think this was a very unwise thing 
to do because it wasted hours of my time to track down and will similarly cost other edgeR users dearly
when their old scripts break. This tool currently now works with 2.4.6.

**Note on prior.N**

http://seqanswers.com/forums/showthread.php?t=5591 says:

*prior.n*

The value for prior.n determines the amount of smoothing of tagwise dispersions towards the common dispersion. 
You can think of it as like a "weight" for the common value. (It is actually the weight for the common likelihood 
in the weighted likelihood equation). The larger the value for prior.n, the more smoothing, i.e. the closer your 
tagwise dispersion estimates will be to the common dispersion. If you use a prior.n of 1, then that gives the 
common likelihood the weight of one observation.

In answer to your question, it is a good thing to squeeze the tagwise dispersions towards a common value, 
or else you will be using very unreliable estimates of the dispersion. I would not recommend using the value that 
you obtained from estimateSmoothing()---this is far too small and would result in virtually no moderation 
(squeezing) of the tagwise dispersions. How many samples do you have in your experiment? 
What is the experimental design? If you have few samples (less than 6) then I would suggest a prior.n of at least 10. 
If you have more samples, then the tagwise dispersion estimates will be more reliable, 
so you could consider using a smaller prior.n, although I would hesitate to use a prior.n less than 5. 


From Bioconductor Digest, Vol 118, Issue 5, Gordon writes:

Dear Dorota,

The important settings are prior.df and trend.

prior.n and prior.df are related through prior.df = prior.n * residual.df,
and your experiment has residual.df = 36 - 12 = 24.  So the old setting of
prior.n=10 is equivalent for your data to prior.df = 240, a very large
value.  Going the other way, the new setting of prior.df=10 is equivalent
to prior.n=10/24.

To recover old results with the current software you would use

  estimateTagwiseDisp(object, prior.df=240, trend="none")

To get the new default from old software you would use

  estimateTagwiseDisp(object, prior.n=10/24, trend=TRUE)

Actually the old trend method is equivalent to trend="loess" in the new
software. You should use plotBCV(object) to see whether a trend is
required.

Note you could also use

  prior.n = getPriorN(object, prior.df=10)

to map between prior.df and prior.n.

----

**Attributions**

edgeR - edgeR_ 

VOOM/limma - limma_VOOM_ 

DESeq2 - DESeq2_ for details

See above for Bioconductor package documentation for packages exposed in Galaxy by this tool and app store package.

Galaxy_ (that's what you are using right now!) for gluing everything together 

Otherwise, all code and documentation comprising this tool was written by Ross Lazarus and is 
licensed to you under the LGPL_ like other rgenetics artefacts

.. _LGPL: http://www.gnu.org/copyleft/lesser.html
.. _HTSeq: http://www-huber.embl.de/users/anders/HTSeq/doc/index.html
.. _edgeR: http://www.bioconductor.org/packages/release/bioc/html/edgeR.html
.. _DESeq2: http://www.bioconductor.org/packages/release/bioc/html/DESeq2.html
.. _limma_VOOM: http://www.bioconductor.org/packages/release/bioc/html/limma.html
.. _Galaxy: http://getgalaxy.org
</help>

</tool>