@@ -12,7 +12,7 @@ Description: Coordinated Gene Activity in Pattern Sets (CoGAPS)

     analysis.

 Maintainer: Elana J. Fertig <ejfertig@jhmi.edu>

 Depends:

-    R (>= 3.0.1),

+    R (>= 3.4.0),

     Rcpp (>= 0.11.0)

 Imports:

     RColorBrewer (>= 1.0.5),

@@ -32,6 +32,7 @@ Suggests:

     testthat,

     lintr

 LinkingTo: Rcpp, BH

+VignetteBuilder: knitr

 License: GPL (==2)

 biocViews: GeneExpression, Transcription, GeneSetEnrichment,

     DifferentialExpression, Bayesian, Clustering, TimeCourse, RNASeq, Microarray,

@@ -23,6 +23,7 @@

 #'  the fixed patterns

 #' @param fixedPatterns matrix of fixed values in either A or P matrix

 #' @param checkpointInterval time (in seconds) between creating a checkpoint

+#' @param ... keeps backwards compatibility with arguments from older versions

 #' @return list with A and P matrix estimates

 #' @importFrom methods new

 #' @export

@@ -55,11 +56,8 @@ fixedPatterns = matrix(0), checkpointInterval=0, ...)

         stop('D and S matrix must be non-negative')

     # run algorithm with call to C++ code

-    nFactor <- floor(nFactor)

-    nEquil <- floor(nEquil)

-    nSample <- floor(nSample)

-    result <- cogaps_cpp(D, S, nFactor, nEquil, floor(nEquil/10), nSample, nOutputs, nSnapshots,

-        alphaA, alphaP, maxGibbmassA, maxGibbmassP, seed, messages,

+    result <- cogaps_cpp(D, S, nFactor, nEquil, nEquil/10, nSample, nOutputs,

+        nSnapshots, alphaA, alphaP, maxGibbmassA, maxGibbmassP, seed, messages,

         singleCellRNASeq, whichMatrixFixed, fixedPatterns, checkpointInterval)

     # backwards compatible with v2

@@ -94,8 +92,22 @@ displayBuildReport <- function()

 #'

 #' @param D data matrix

 #' @param S uncertainty matrix

+#' @param ABins unused

+#' @param PBins unused

+#' @param simulation_id unused

+#' @param nOutR number of output messages

+#' @param output_atomic unused

+#' @param fixedBinProbs unused

+#' @param fixedDomain unused

+#' @param sampleSnapshots indicates if snapshots should be made

+#' @param numSnapshots how many snapshots to take

+#' @param nMaxA unused

+#' @param nMaxP unused

+#' @param max_gibbmass_paraA limit truncated normal to max size

+#' @param max_gibbmass_paraP limit truncated normal to max size

 #' @return list with A and P matrix estimates

 #' @importFrom methods new

+#' @inheritParams CoGAPS

 #' @export

 gapsRun <- function(D, S, ABins=data.frame(), PBins=data.frame(), nFactor=7,

 simulation_id="simulation", nEquil=1000, nSample=1000, nOutR=1000,

@@ -115,9 +127,11 @@ alphaP=0.01, nMaxP=100000, max_gibbmass_paraP=100.0, seed=-1, messages=TRUE)

 #' @param D data matrix

 #' @param S uncertainty matrix

 #' @param FP data.frame with rows giving fixed patterns for P

+#' @param fixedMatrix unused

 #' @param ... v2 style parameters

 #' @return list with A and P matrix estimates

 #' @importFrom methods new

+#' @inheritParams gapsRun

 #' @export

 gapsMapRun <- function(D, S, FP, ABins=data.frame(), PBins=data.frame(),

 nFactor=5, simulation_id="simulation", nEquil=1000, nSample=1000, nOutR=1000,

@@ -34,7 +34,7 @@ cluster.method="complete", ignore.NA=FALSE, bySet=FALSE, ...)

     corr.dist=1-corr.dist

     # cluster

     #library(cluster)

-    clust=agnes(x=corr.dist,diss=T,method=cluster.method)

+    clust=agnes(x=corr.dist,diss=TRUE,method=cluster.method)

     cut=cutree(as.hclust(clust),k=cnt)

     #save.image(file=paste("CoGAPS.",nP,"P.",nS,"Set.CorrClustCut",cnt,".RData"))

@@ -1,17 +1,21 @@

 #'\code{plotSmoothPatterns} plots the output A and P matrices as a

 #' heatmap and line plot respectively

 #'

-#'@param P the mean A matrix

-#'@param x optional variables

-#'@param breaks breaks in plots

-#'@param breakStyle style of breaks

-#'@param orderP whether to order patterns

-#'@param plotPTS whether to plot points on lines

-#'@param pointCol color of points

-#'@param lineCol color of line

-#'@param add logical specifying if bars should be added to an already existing plot; defaults to `FALSE'.

-#'@param ... arguments to be passed to/from other methods.  For the default method these can include further arguments (such as `axes', `asp' and `main') and graphical parameters (see `par') which are passed to `plot.window()', `title()' and `axis'.

-#'@export

+#' @param P the mean A matrix

+#' @param x optional variables

+#' @param breaks breaks in plots

+#' @param breakStyle style of breaks

+#' @param orderP whether to order patterns

+#' @param plotPTS whether to plot points on lines

+#' @param pointCol color of points

+#' @param lineCol color of line

+#' @param add logical specifying if bars should be added to an already existing

+#'  plot; defaults to `FALSE'.

+#' @param ... arguments to be passed to/from other methods.  For the default

+#'  method these can include further arguments (such as `axes', `asp' and

+#'  `main') and graphical parameters (see `par') which are passed to

+#"  `plot.window()', `title()' and `axis'.

+#' @export

 plotSmoothPatterns <- function(P, x=NULL, breaks=NULL, breakStyle=TRUE,

 orderP=!all(is.null(x)), plotPTS=FALSE, pointCol='black', lineCol='grey',

 add=FALSE, ...)

@@ -31,7 +35,8 @@ add=FALSE, ...)

     # make the breaks in a uniform format

     if (length(breaks)==1)

     {

-        breaks <- as.numeric(unique(unlist(strsplit(sub("\\(","",sub("\\]","",levels(cut(x,breaks)))),split=","))))

+        breaks <- as.numeric(unique(unlist(strsplit(sub("\\(","",sub("\\]","",

+            levels(cut(x,breaks)))),split=","))))

     }

     # check that the style of breaks matches the number of groups

@@ -67,7 +72,7 @@ add=FALSE, ...)

         xMax <- seq(from=ncol(P)+1,length.out=nrow(P))[order(PMax,decreasing=TRUE)]

         xTmp <- x

         PTmp <- P

-        for (iP in order(PMax,decreasing=T))

+        for (iP in order(PMax,decreasing=TRUE))

         {

             if (length(xTmp) < 1)

             {

@@ -92,7 +97,7 @@ add=FALSE, ...)

     }

     else

     {

-        split.screen(c(nrow(P), length(which(breakStyle))), erase=F)

+        split.screen(c(nrow(P), length(which(breakStyle))), erase=FALSE)

     }

     scr <- 1

@@ -119,7 +124,7 @@ add=FALSE, ...)

             if (add)

             {

-                plot(x[idxTmp], loess(P[iP,idxTmp]~x[idxTmp])$fit, col=lineCol, type='l', axes=F, xlab='', ylab='',lwd=3,

+                plot(x[idxTmp], loess(P[iP,idxTmp]~x[idxTmp])$fit, col=lineCol, type='l', axes=FALSE, xlab='', ylab='',lwd=3,

                     ylim=c(0,max(P[iP,])),xlim=xBorders,...)

             }

             else

@@ -47,6 +47,8 @@ the fixed patterns}

 \item{fixedPatterns}{matrix of fixed values in either A or P matrix}

 \item{checkpointInterval}{time (in seconds) between creating a checkpoint}

+

+\item{...}{keeps backwards compatibility with arguments from older versions}

 }

 \value{

 list with A and P matrix estimates

@@ -19,6 +19,50 @@ gapsMapRun(D, S, FP, ABins = data.frame(), PBins = data.frame(),

 \item{FP}{data.frame with rows giving fixed patterns for P}

+\item{ABins}{unused}

+

+\item{PBins}{unused}

+

+\item{nFactor}{number of patterns (basis vectors, metagenes), which must be

+greater than or equal to the number of rows of FP}

+

+\item{simulation_id}{unused}

+

+\item{nEquil}{number of iterations for burn-in}

+

+\item{nSample}{number of iterations for sampling}

+

+\item{nOutR}{number of output messages}

+

+\item{output_atomic}{unused}

+

+\item{fixedMatrix}{unused}

+

+\item{fixedBinProbs}{unused}

+

+\item{fixedDomain}{unused}

+

+\item{sampleSnapshots}{indicates if snapshots should be made}

+

+\item{numSnapshots}{how many snapshots to take}

+

+\item{alphaA}{sparsity parameter for A domain}

+

+\item{nMaxA}{unused}

+

+\item{max_gibbmass_paraA}{limit truncated normal to max size}

+

+\item{alphaP}{sparsity parameter for P domain}

+

+\item{nMaxP}{unused}

+

+\item{max_gibbmass_paraP}{limit truncated normal to max size}

+

+\item{seed}{a positive seed is used as-is, while any negative seed tells

+the algorithm to pick a seed based on the current time}

+

+\item{messages}{display progress messages}

+

 \item{...}{v2 style parameters}

 }

 \value{

@@ -15,6 +15,48 @@ gapsRun(D, S, ABins = data.frame(), PBins = data.frame(), nFactor = 7,

 \item{D}{data matrix}

 \item{S}{uncertainty matrix}

+

+\item{ABins}{unused}

+

+\item{PBins}{unused}

+

+\item{nFactor}{number of patterns (basis vectors, metagenes), which must be

+greater than or equal to the number of rows of FP}

+

+\item{simulation_id}{unused}

+

+\item{nEquil}{number of iterations for burn-in}

+

+\item{nSample}{number of iterations for sampling}

+

+\item{nOutR}{number of output messages}

+

+\item{output_atomic}{unused}

+

+\item{fixedBinProbs}{unused}

+

+\item{fixedDomain}{unused}

+

+\item{sampleSnapshots}{indicates if snapshots should be made}

+

+\item{numSnapshots}{how many snapshots to take}

+

+\item{alphaA}{sparsity parameter for A domain}

+

+\item{nMaxA}{unused}

+

+\item{max_gibbmass_paraA}{limit truncated normal to max size}

+

+\item{alphaP}{sparsity parameter for P domain}

+

+\item{nMaxP}{unused}

+

+\item{max_gibbmass_paraP}{limit truncated normal to max size}

+

+\item{seed}{a positive seed is used as-is, while any negative seed tells

+the algorithm to pick a seed based on the current time}

+

+\item{messages}{display progress messages}

 }

 \value{

 list with A and P matrix estimates

@@ -26,9 +26,12 @@ plotSmoothPatterns(P, x = NULL, breaks = NULL, breakStyle = TRUE,

 \item{lineCol}{color of line}

-\item{add}{logical specifying if bars should be added to an already existing plot; defaults to `FALSE'.}

+\item{add}{logical specifying if bars should be added to an already existing

+plot; defaults to `FALSE'.}

-\item{...}{arguments to be passed to/from other methods.  For the default method these can include further arguments (such as `axes', `asp' and `main') and graphical parameters (see `par') which are passed to `plot.window()', `title()' and `axis'.}

+\item{...}{arguments to be passed to/from other methods.  For the default

+method these can include further arguments (such as `axes', `asp' and

+`main') and graphical parameters (see `par') which are passed to}

 }

 \description{

 \code{plotSmoothPatterns} plots the output A and P matrices as a

@@ -1,13 +1,15 @@

 ---

 title: "GWCoGAPS and PatternMarkers Vignette"

 author: "Genevieve L. Stein-O'Brien"

-date: \today

-output: BiocStyle::pdf_document

+date: "`r doc_date()`"

+package: "`r pkg_ver('CoGAPS')`"

 bibliography: AppNote.bib

 vignette: >

-  %\VignetteIndexEntry{Overview of CNPBayes package}

+  %\VignetteIndexEntry{GWCoGAPS and PatternMarkers}

   %\VignetteEngine{knitr::rmarkdown}

-  %\usepackage[utf8]{inputenc} 

+  %\VignetteEncoding{UTF-8}

+output: 

+  BiocStyle::html_document

 ---

 # Introduction

@@ -25,11 +27,13 @@ pipeline for genome wide NMF analysis.

 The GWCoGAPS algorithm is run by calling the GWCoGAPS function in the CoGAPS

 R package as follows:

-```{r, eval=FALSE}

-library(CoGAPS) 

+```{r}

+library(CoGAPS)

+```

+***

 GWCoGAPS(D, S, nFactor, nSets, nCores, saveBySetResults, fname,

     PatternsMatchFN = patternMatch4Parallel, Cut, minNS, ...)

-```

+***

 **Input Arguments**

 The inputs that must be set each time are only the nSets, nFactor, and data and standard deviation matrices, with all other inputs having default values. Additional inputs to the gapsRun function, as outlined previously, can be used to taylor the analysis based on the expected dimensionality of the data. GWCoGAPS specific arguments are as follows:

@@ -54,14 +58,15 @@ greater than or equal to the number of rows of FP}

 \par In this example, we use the same simulated data in SimpSim (SimpSim.D), as previously described, with three known patterns (SimpSim.P) and corresponding amplitude (SimpSim.A) with specified activity in two gene sets (GSets).

-```{r , eval=FALSE}

+***

 library('CoGAPS')

 data('SimpSim')

 GWCoGAPS(SimpSim.D, SimpSim.S, nFactor=3, nCores=NA, nSets=2,

             fname="test" ,PatternsMatchFN = postCoGAPSPatternMatch,

             sampleSnapshots = "TRUE", numSnapshots = 3)

 plotGAPS(AP.Fixed$A, AP.Fixed$P, 'ModSimFigs')

-```

+***

+

 Figure \ref{fig:ModSim} shows the results from plotting the GWCoGAPS estimates of ${\bf{A}}$ and ${\bf{P}}$ using \texttt{plotGAPS}.  

 \begin{figure}[ht]

   \begin{center}

@@ -113,11 +118,14 @@ Parallelizing the GAPS using the doParallel package returns a list containing th

 \item[PSnapshots]{Samples of P matrices taken during sampling.}

 \end{description}

 To ease accross set pattern comparisons, the reOrderBySet function restructures this list such that all nSets sets solution for Amean, Pmean, and Asd are listed under ${\bf{A}} ,{\bf{P}}$, and ${\bf{Asd}}$, respectively. NMF solutions using the same syntax can also be imput to reOrderBySet via the AP arguement as follows:

-```{r, eval=FALSE}

+

+***

 BySet<-reOrderBySet(AP=AP,nFactor=nFactor,nSets=nSets)

-```

+***

+

 The resulting list of pattern matrixes, i.e. BySet$P, can serve as input for the patternMatch4Parallel function or further transformed for alternative accross set pattern matching techniques. Manual curration of pattern matching can be accomplished by iterating between the patternMatch4Parallel function and visual editting using the PatternMatcher as follows: 

-```{r , eval=FALSE}

+

+***

 matchedPs<-patternMatch4Parallel(Ptot=BySet$P,nP=nFactor,nS=nSets,

                                  cnt=nFactor,minNS=minNS,bySet=TRUE)

 selectPBySet<-PatternMatcher(PBySet=matchedPs[["PBySet"]])

@@ -127,13 +135,14 @@ rownames(selectPBySet)<-selectPBySet$BySet

 selectPBySet<-as.matrix(selectPBySet[,-1])

 matchedPs<-patternMatch4Parallel(Ptot=selectPBySet,nP=nFactor,nS=nSets,

                                  cnt=nFactor,minNS=minNS,bySet=FALSE)

-```

+***

+

 The output of this process, matchedPs, is a data.frame of consensus patterns which can be directly read into the FP arguement of gapsMapRun. These patterns are scaled to have a maximum of one to allow for direct accross pattern comparisions of the ${\bf{A}}$ values both within and accross parallel sets. 

 ### Example: Simulated data

 \par This example will manually generate the same result as calling the GWCoGAPS function as given in the example in the previous section.

-```{r , eval=FALSE}

+***

 data('SimpSim')

 D<-SimpSim.D

 S<-SimpSim.S

@@ -190,8 +199,7 @@ Fixed <- foreach(i=1:nSets) %dopar% {

 #extract A and Asds

 As4fixPs <- postFixed(AP.fixed=Fixed,setPs=matchedPs)

-```

-

+***

 # PatternMarkers

@@ -212,10 +220,10 @@ where  are the elements of the $\bf{A}$ matrix for the $\textit{i}^{th}$ gene sc

 The PatternMarkers statistic is run by calling thepatternMarkers function in the CoGAPS R package as follows:

-```{r , eval=FALSE}

+***

  patternMarkers(Amatrix = AP$Amean, scaledPmatrix = FALSE, Pmatrix = NA,

   threshold = "All", lp = NA, full = FALSE, ...)

-```

+***

 **Input Arguments**

 \begin{description}

@@ -229,11 +237,11 @@ The PatternMarkers statistic is run by calling thepatternMarkers function in the

 Once the PatternMarkers statistic has been run, a heatmap of each markers expression level can be displayed using the plotPatternMarkers function as follows:

-```{r , eval=FALSE}

+***

 plotPatternMarkers(data = NA, patternMarkers = PatternMarkers,

       patternPalette = NA, sampleNames = NA, samplePalette = NA,

       colDenogram = TRUE, heatmapCol = "bluered", scale = "row", ...)

-```

+***

 **Input Arguments**

 \begin{description}

@@ -246,10 +254,10 @@ plotPatternMarkers(data = NA, patternMarkers = PatternMarkers,

 ### Example: Simulated data

-```{r , eval=FALSE}

+***

 PatternMarkers<-patternMarkers(Amatrix=AP.fixed$A,scaledPmatrix=TRUE,threshold="cut")

 plotPatternMarkers(data=D,patternMarkers=PatternMarkers,patternPalette=c("grey","navy","orange"))

-```

+***

 Figure \ref{fig:PM1} shows the results from running plotPatternMarkers on the PatternMarkers generated from the the GWCoGAPS results generated from the simulated data as previously illustrated.

 \begin{figure}[h]