miga-base 0.3.9.0 → 0.3.9.1

data/utils/enveomics/enveomics.R/R/recplot2.R

@@ -9,6 +9,7 @@ setClass("enve.RecPlot2",
    id.counts='numeric',		##<< Counts per ID bin.
    id.breaks='numeric',		##<< Breaks of identity bins.
    pos.breaks='numeric',	##<< Breaks of position bins.
+   pos.names='character',       ##<< Names of the position bins.
    seq.breaks='numeric',	##<< Breaks of input sequences.
    peaks='list',                ##<< Peaks identified in the recplot.
    ### Limits of the subject sequences after concatenation.
@@ -70,10 +71,17 @@ plot.enve.RecPlot2 <- function
       ###   3: identity histogram,
       ###   4: Populations histogram (histogram of sequencing depths),
       ###   5: Color scale for the counts matrix (vertical),
-      ###   6: Color scale of the counts
-      ### matrix (horizontal). Only panels indicated here will be plotted. To
-      ### plot only one panel simply set this to the number of the panel you
-      ### want to plot.
+      ###   6: Color scale of the counts matrix (horizontal)
+      ### Only panels indicated here will be plotted. To plot only one panel
+      ### simply set this to the number of the panel you want to plot.
+      panel.fun=list(),
+      ### List of functions to be executed after drawing each panel. Use the
+      ### indices in `layout` (as characters) as keys. Functions for indices
+      ### missing in `layout` are ignored. For example, to add a vertical line
+      ### at the 3Mbp mark in both the position histogram and the counts matrix:
+      ### `list('1'=function() abline(v=3), '2'=function() abline(v=3))`.
+      ### Note that the X-axis in both panels is in Mbp by default. To change
+      ### this behavior, set `pos.units` accordingly.
       widths=c(1,7,2),
       ### Relative widths of the columns of `layout`.
       heights=c(1,2),
@@ -166,7 +174,7 @@ plot.enve.RecPlot2 <- function
 		  list(maxColorValue=256, alpha=52)));
    }
 
-   # Counts matrix
+   # [1] Counts matrix
    if(any(layout==1)){
       par(mar=mar[['1']]);
       plot(1, t='n', bty='l',
@@ -182,9 +190,10 @@ plot.enve.RecPlot2 <- function
       image(x=pos.breaks, y=id.breaks, z=log10(counts),col=palette,
 	 bg=grey(1,0), breaks=seq(-.1,log10(max(counts)),
 	 length.out=1+length(palette)), add=TRUE);
+      if(exists('1',panel.fun)) panel.fun[['1']]();
    }
 
-   # Position histogram
+   # [2] Position histogram
    if(any(layout==2)){
       par(mar=mar[['2']]);
       if(any(layout==1)){
@@ -213,9 +222,10 @@ plot.enve.RecPlot2 <- function
       if(any(pos.counts.in==0))  rect(pos.breaks[c(pos.counts.in==0,FALSE)],
 	       seqdepth.lim[1], pos.breaks[c(FALSE,pos.counts.in==0)],
 	       seqdepth.lim[1]*3/2, col=in.col,  border=NA);
+      if(exists('2',panel.fun)) panel.fun[['2']]();
    }
 
-   # Identity histogram
+   # [3] Identity histogram
    if(any(layout==3)){
       par(mar=mar[['3']]);
       if(any(layout==1)){
@@ -248,9 +258,10 @@ plot.enve.RecPlot2 <- function
 	 plot(1,t='n',bty='l',xlab='', xaxt='n', ylab='', yaxt='n')
 	 text(1,1,labels='Insufficient data', srt=90)
       }
+      if(exists('3',panel.fun)) panel.fun[['3']]();
    }
 
-   # Populations histogram
+   # [4] Populations histogram
    peaks <- NA;
    if(any(layout==4)){
       par(mar=mar[['4']]);
@@ -308,9 +319,10 @@ plot.enve.RecPlot2 <- function
                     dpt,'X (', frx, '%', err, ')', sep=''))
 	 }
       }
+      if(exists('4',panel.fun)) panel.fun[['4']]();
    }
 
-   # Color scale
+   # [5] Color scale of the counts matrix (vertical)
    count.bins <- 10^seq(log10(min(counts[counts>0])), log10(max(counts)),
       length.out=1+length(palette))
    if(any(layout==5)){
@@ -319,13 +331,17 @@ plot.enve.RecPlot2 <- function
 	 ylim=range(count.bins), yaxs='i', ylab='')
       rect(0,count.bins[-length(count.bins)],1,count.bins[-1],col=palette,
 	 border=NA)
+      if(exists('5',panel.fun)) panel.fun[['5']]();
    }
+
+   # [6] Color scale of the coutnts matrix (horizontal)
    if(any(layout==6)){
       par(mar=mar[['6']]);
       plot(1,t='n',log='x',ylim=0:1,yaxt='n',ylab='',yaxs='i',
 	 xlim=range(count.bins), xaxs='i',xlab='');
       rect(count.bins[-length(count.bins)],0,count.bins[-1],1,col=palette,
 	 border=NA);
+      if(exists('6',panel.fun)) panel.fun[['6']]();
    }
    
    par(mar=ori.mar);
@@ -337,113 +353,129 @@ plot.enve.RecPlot2 <- function
 
 #==============> Define core functions
 enve.recplot2 <- function(
-   ### Produces recruitment plots provided that BlastTab.catsbj.pl has
-   ### been previously executed.
-      prefix,
-      ### Path to the prefix of the BlastTab.catsbj.pl output files. At
-      ### least the files .rec and .lim must exist with this prefix.
-      plot=TRUE,
-      ### Should the object be plotted?
-      pos.breaks=1e3,
-      ### Breaks in the positions histogram. It can also be a vector of break
-      ### points, and values outside the range are ignored. If zero (0), it
-      ### uses the sequence breaks as defined in the .lim file, which means
-      ### one bin per contig (or gene, if the mapping is agains genes).
-      id.breaks=300,
-      ### Breaks in the identity histogram. It can also be a vector of break
-      ### points, and values outside the range are ignored.
-      id.free.range=FALSE,
-      ### Indicates that the range should be freely set from the observed
-      ### values. Otherwise, 70-100% is included in the identity histogram
-      ### (default).
-      id.metric=c('identity', 'corrected identity', 'bit score'),
-      ### Metric of identity to be used (Y-axis). Corrected identity is only
-      ### supported if the original BLAST file included sequence lengths.
-      id.summary=sum,
-      ### Function summarizing the identity bins. Other recommended options
-      ### include: `median` to estimate the median instead of total bins, and
-      ### `function(x) mlv(x,method='parzen')$M` to estimate the mode.
-      id.cutoff=95,
-      ### Cutoff of identity metric above which the hits are considered
-      ### 'in-group'. The 95% identity corresponds to the expectation of
-      ### ANI<95% within species.
-      threads=2,
-      ### Number of threads to use.
-      verbose=TRUE,
-      ### Indicates if the function should report the advance.
-      ...
-      ### Any additional parameters supported by `plot.enve.RecPlot2`.
-   ){
-   # Settings
-   id.metric <- match.arg(id.metric);
-   
-   #Read files
-   if(verbose) cat("Reading files.\n")
-   rec <- read.table(paste(prefix, ".rec", sep=""), sep="\t", comment.char="",
-      quote="");
-   lim <- read.table(paste(prefix, ".lim", sep=""), sep="\t", comment.char="",
-      quote="", as.is=TRUE);
-   
-   # Build matrix
-   if(verbose) cat("Building counts matrix.\n")
-   if(id.metric=="corrected identity" & ncol(rec)<6){
-      stop("Requesting corr. identity, but .rec file doesn't have 6th column")
-   }
-   rec.idcol <- ifelse(id.metric=="identity", 3,
-      ifelse(id.metric=="corrected identity", 6, 4));
-   if(length(pos.breaks)==1){
-      if(pos.breaks>0){
-         pos.breaks <- seq(min(lim[,2]), max(lim[,3]), length.out=pos.breaks+1);
-      }else{
-         pos.breaks <- c(lim[1,2], lim[,3])
-      }
-   }
-   if(length(id.breaks)==1){
-      id.range.v <- rec[,rec.idcol]
-      if(!id.free.range) id.range.v <- c(id.range.v,70,100)
-      id.range.v <- range(id.range.v)
-      id.breaks <- seq(id.range.v[1], id.range.v[2], length.out=id.breaks+1);
-   }
-   
-   # Run in parallel
-   if(nrow(rec) < 200) threads <- 1 # It doesn't worth the overhead
-   cl		<- makeCluster(threads)
-   rec.l	<- list()
-   thl		<- ceiling(nrow(rec)/threads)
-   for(i in 0:(threads-1)){
-      rec.l[[i+1]] <- list(rec=rec[ (i*thl+1):min(((i+1)*thl),nrow(rec)), ],
-			verbose=ifelse(i==0, verbose, FALSE))
-   }
-   counts.l	<- clusterApply(cl, rec.l, enve.recplot2.__counts,
-			pos.breaks=pos.breaks, id.breaks=id.breaks,
-			rec.idcol=rec.idcol)
-   counts	<- counts.l[[1]]
-   if(threads>1) for(i in 2:threads) counts <- counts + counts.l[[i]]
-   stopCluster(cl)
-   
-   # Estimate 1D histograms
-   if(verbose) cat("Building histograms.\n")
-   id.mids	<- (id.breaks[-length(id.breaks)]+id.breaks[-1])/2;
-   id.ingroup	<- (id.mids > id.cutoff);
-   id.counts	<- apply(counts, 2, id.summary);
-   pos.counts.in   <- apply(counts[,id.ingroup], 1, sum);
-   pos.counts.out  <- apply(counts[,!id.ingroup], 1, sum);
+  ### Produces recruitment plots provided that BlastTab.catsbj.pl has
+  ### been previously executed.
+    prefix,
+    ### Path to the prefix of the BlastTab.catsbj.pl output files. At
+    ### least the files .rec and .lim must exist with this prefix.
+    plot=TRUE,
+    ### Should the object be plotted?
+    pos.breaks=1e3,
+    ### Breaks in the positions histogram. It can also be a vector of break
+    ### points, and values outside the range are ignored. If zero (0), it
+    ### uses the sequence breaks as defined in the .lim file, which means
+    ### one bin per contig (or gene, if the mapping is agains genes). Ignored
+    ### if `pos.breaks.tsv` is passed.
+    pos.breaks.tsv=NA,
+    ### Path to a list of (absolute) coordinates to use as position breaks.
+    ### This tab-delimited file can be produced by `GFF.catsbj.pl`, and it
+    ### must contain at least one column: coordinates of the break positions of
+    ### each position bin. If it has a second column, this is used as the name
+    ### of the position bin that ends at the given coordinate (the first row is
+    ### ignored). Any additional columns are currently ignored. If NA,
+    ### position bins are determined by `pos.breaks`.
+    id.breaks=300,
+    ### Breaks in the identity histogram. It can also be a vector of break
+    ### points, and values outside the range are ignored.
+    id.free.range=FALSE,
+    ### Indicates that the range should be freely set from the observed
+    ### values. Otherwise, 70-100% is included in the identity histogram
+    ### (default).
+    id.metric=c('identity', 'corrected identity', 'bit score'),
+    ### Metric of identity to be used (Y-axis). Corrected identity is only
+    ### supported if the original BLAST file included sequence lengths.
+    id.summary=sum,
+    ### Function summarizing the identity bins. Other recommended options
+    ### include: `median` to estimate the median instead of total bins, and
+    ### `function(x) mlv(x,method='parzen')$M` to estimate the mode.
+    id.cutoff=95,
+    ### Cutoff of identity metric above which the hits are considered
+    ### 'in-group'. The 95% identity corresponds to the expectation of
+    ### ANI<95% within species.
+    threads=2,
+    ### Number of threads to use.
+    verbose=TRUE,
+    ### Indicates if the function should report the advance.
+    ...
+    ### Any additional parameters supported by `plot.enve.RecPlot2`.
+  ){
+  # Settings
+  id.metric <- match.arg(id.metric);
 
-   # Plot and return
-   recplot <- new('enve.RecPlot2',
-      counts=counts, id.counts=id.counts, pos.counts.in=pos.counts.in,
-      pos.counts.out=pos.counts.out,
-      id.breaks=id.breaks, pos.breaks=pos.breaks,
-      seq.breaks=c(lim[1,2], lim[,3]), seq.names=lim[,1],
-      id.ingroup=id.ingroup,id.metric=id.metric,
-      call=match.call());
-   if(plot){
-      if(verbose) cat("Plotting.\n")
-      peaks <- plot(recplot, ...);
-      attr(recplot, "peaks") <- peaks
-   }
-   return(recplot);
-   ### Returns an object of class `enve.RecPlot2`.
+  #Read files
+  if(verbose) cat("Reading files.\n")
+  rec <- read.table(paste(prefix, ".rec", sep=""), sep="\t", comment.char="",
+        quote="");
+  lim <- read.table(paste(prefix, ".lim", sep=""), sep="\t", comment.char="",
+        quote="", as.is=TRUE);
+
+  # Build matrix
+  if(verbose) cat("Building counts matrix.\n")
+  if(id.metric=="corrected identity" & ncol(rec)<6){
+    stop("Requesting corr. identity, but .rec file doesn't have 6th column")
+  }
+  rec.idcol <- ifelse(id.metric=="identity", 3,
+        ifelse(id.metric=="corrected identity", 6, 4))
+  pos.names <- as.character(NULL)
+  if(!is.na(pos.breaks.tsv)){
+    tmp <- read.table(pos.breaks.tsv, sep='\t', header=FALSE, as.is=TRUE)
+    pos.breaks <- as.numeric(tmp[,1])
+    if(ncol(tmp)>1) pos.names <- as.character(tmp[-1,2])
+  }else if(length(pos.breaks)==1){
+    if(pos.breaks>0){
+      pos.breaks <- seq(min(lim[,2]), max(lim[,3]), length.out=pos.breaks+1);
+    }else{
+      pos.breaks <- c(lim[1,2], lim[,3])
+      pos.names  <- lim[,1]
+    }
+  }
+  if(length(id.breaks)==1){
+    id.range.v <- rec[,rec.idcol]
+    if(!id.free.range) id.range.v <- c(id.range.v,70,100)
+    id.range.v <- range(id.range.v)
+    id.breaks <- seq(id.range.v[1], id.range.v[2], length.out=id.breaks+1);
+  }
+
+  # Run in parallel
+  if(nrow(rec) < 200) threads <- 1 # It doesn't worth the overhead
+  cl    <- makeCluster(threads)
+  rec.l <- list()
+  thl   <- ceiling(nrow(rec)/threads)
+  for(i in 0:(threads-1)){
+    rec.l[[i+1]] <- list(
+          rec=rec[ (i*thl+1):min(((i+1)*thl),nrow(rec)), ],
+          verbose=ifelse(i==0, verbose, FALSE))
+  }
+  counts.l <- clusterApply(cl, rec.l, enve.recplot2.__counts,
+                pos.breaks=pos.breaks, id.breaks=id.breaks,
+                rec.idcol=rec.idcol)
+  counts   <- counts.l[[1]]
+  if(threads>1) for(i in 2:threads) counts <- counts + counts.l[[i]]
+  stopCluster(cl)
+
+  # Estimate 1D histograms
+  if(verbose) cat("Building histograms.\n")
+  id.mids	<- (id.breaks[-length(id.breaks)]+id.breaks[-1])/2;
+  id.ingroup	<- (id.mids > id.cutoff);
+  id.counts	<- apply(counts, 2, id.summary);
+  pos.counts.in   <- apply(counts[,id.ingroup], 1, sum);
+  pos.counts.out  <- apply(counts[,!id.ingroup], 1, sum);
+
+  # Plot and return
+  recplot <- new('enve.RecPlot2',
+    counts=counts, id.counts=id.counts, pos.counts.in=pos.counts.in,
+    pos.counts.out=pos.counts.out,
+    id.breaks=id.breaks, pos.breaks=pos.breaks, pos.names=pos.names,
+    seq.breaks=c(lim[1,2], lim[,3]), seq.names=lim[,1],
+    id.ingroup=id.ingroup,id.metric=id.metric,
+    call=match.call());
+  if(plot){
+    if(verbose) cat("Plotting.\n")
+    peaks <- plot(recplot, ...);
+    attr(recplot, "peaks") <- peaks
+  }
+  return(recplot);
+  ### Returns an object of class `enve.RecPlot2`.
 }
 
 enve.recplot2.findPeaks <- function(
@@ -502,9 +534,11 @@ enve.recplot2.findPeaks.emauto <- function(
     stop('Invalid criterion ', criterion)
   }
   for(comp in components){
+    if(verbose) cat('Testing:',comp,'\n')
     best <- enve.recplot2.findPeaks.__emauto_one(x, comp, do_crit, best,
           verbose, ...)
   }
+  if(length(best[['peaks']])==0) return(list())
 
   seqdepths.r <- signif(log(sapply(best[['peaks']],
         function(x) x$seq.depth)), merge.tol)
@@ -609,7 +643,7 @@ enve.recplot2.findPeaks.mower <- function(
       ### Range of quantiles to be used in the estimation of a peak's
       ### parameters.
       mlv.opts=list(method='parzen'),
-      ### Options passed to `mlv` to estimate the mode.
+      ### Ignored. For backwards compatibility.
       fitdist.opts.sn=list(distr='sn', method='qme', probs=c(0.1,0.5,0.8),
 	 start=list(omega=1, alpha=-1), lower=c(0, -Inf, -Inf)),
       ### Options passed to `fitdist` to estimate the standard deviation if
@@ -758,53 +792,75 @@ enve.recplot2.changeCutoff <- function
    return(rp)
 }
 
-enve.recplot2.extractWindows <- function
-   ### Extract windows significantly below (or above) the peak in sequencing
-   ### depth.
-      (rp,
-      ### Recruitment plot, a enve.RecPlot2 object.
-      peak,
-      ### Peak, an `enve.RecPlot2.Peak` object. If list, it is assumed to be a
-      ### list of enve.RecPlot2.Peak objects, in which case the core peak is
-      ### used (see `enve.recplot2.corePeak`).
-      lower.tail=TRUE,
-      ### If FALSE, it returns windows significantly above the peak in
-      ### sequencing depth.
-      significance=0.05,
-      ### Significance threshold (alpha) to select windows.
-      seq.names=FALSE
-      ### Returns subject sequence names instead of a vector of Booleans. If
-      ### the recruitment plot was generated with pos.breaks=0 it returns a
-      ### vector of characters (the sequence identifiers), otherwise it returns
-      ### a data.frame with a name column and two columns of coordinates.
-      ){
-   # Determine the threshold
+enve.recplot2.windowDepthThreshold <- function
+  ### Identifies the threshold below which windows should be identified as
+  ### variable or absent.
+    (rp,
+    ### Recruitment plot, an `enve.RecPlot2` object.
+    peak,
+    ### Peak, an `enve.RecPlot2.Peak` object. If list, it is assumed to be a
+    ### list of `enve.RecPlot2.Peak` objects, in which case the core peak is
+    ### used (see `enve.recplot2.corePeak`).
+    lower.tail=TRUE,
+    ### If FALSE, it returns windows significantly above the peak in
+    ### sequencing depth.
+    significance=0.05
+    ### Significance threshold (alpha) to select windows.
+    ){
    if(is.list(peak)) peak <- enve.recplot2.corePeak(peak)
    par <- peak$param.hat
    par[["p"]] <- ifelse(lower.tail, significance, 1-significance)
    thr <- do.call(ifelse(length(par)==4, qsn, qnorm), par)
    if(peak$log) thr <- exp(thr)
-   
-   # Select windows past the threshold
-   seqdepth.in <- enve.recplot2.seqdepth(rp)
-   if(lower.tail){
-      sel <- seqdepth.in < thr
-   }else{
-      sel <- seqdepth.in > thr
-   }
-   
-   # seq.names=FALSE
-   if(!seq.names) return(sel)
-   # seq.names=TRUE and pos.breaks=0
-   if(length(rp$pos.breaks)==length(rp$seq.breaks) &&
-         rp$pos.breaks==rp$seq.breaks)
-           return(rp$seq.names[sel])
-   # seq.names=TRUE and pos.breaks!=0
-   return(enve.recplot2.coordinates(rp,sel))
-   ### Returns a vector of logicals if `seq.names=FALSE`. If `seq.names=TRUE`,
-   ### it returns a vector of characters if the object was built with
-   ### `pos.breaks=0` or a data.frame with four columns otherwise: name.from,
-   ### name.to, pos.from, and pos.to (see `enve.recplot2.coordinates`).
+
+   return(thr)
+   ### Returns a float. The units are depth if the peaks were estimated in
+   ### linear scale, or log-depth otherwise (`peak$log`).
+}
+
+enve.recplot2.extractWindows <- function
+  ### Extract windows significantly below (or above) the peak in sequencing
+  ### depth.
+    (rp,
+    ### Recruitment plot, a `enve.RecPlot2` object.
+    peak,
+    ### Peak, an `enve.RecPlot2.Peak` object. If list, it is assumed to be a
+    ### list of `enve.RecPlot2.Peak` objects, in which case the core peak is
+    ### used (see `enve.recplot2.corePeak`).
+    lower.tail=TRUE,
+    ### If FALSE, it returns windows significantly above the peak in
+    ### sequencing depth.
+    significance=0.05,
+    ### Significance threshold (alpha) to select windows.
+    seq.names=FALSE
+    ### Returns subject sequence names instead of a vector of Booleans. If
+    ### the recruitment plot was generated with named position bins (e.g, using
+    ### `pos.breaks`=0 or a two-column `pos.breaks.tsv`), it returns a vector of
+    ### characters (the sequence identifiers), otherwise it returns a data.frame
+    ### with a name column and two columns of coordinates.
+  ){
+  # Determine the threshold
+  thr <- enve.recplot2.windowDepthThreshold(rp, peak, lower.tail, significance)
+
+  # Select windows past the threshold
+  seqdepth.in <- enve.recplot2.seqdepth(rp)
+  if(lower.tail){
+    sel <- seqdepth.in < thr
+  }else{
+    sel <- seqdepth.in > thr
+  }
+
+  # seq.names=FALSE
+  if(!seq.names) return(sel)
+  # seq.names=TRUE and pos.names defined
+  if(length(rp$pos.names) != 0) return(rp$pos.names[sel])
+  # seq.names=TRUE and pos.names undefined
+  return(enve.recplot2.coordinates(rp,sel))
+  ### Returns a vector of logicals if `seq.names=FALSE`. If `seq.names=TRUE`,
+  ### it returns a vector of characters if the object has `pos.names` defined,
+  ### or a data.frame with four columns otherwise:
+  ### name.from, name.to, pos.from, and pos.to
+  ### (see `enve.recplot2.coordinates`).
 }
 
 enve.recplot2.compareIdentities <- function
@@ -931,7 +987,11 @@ enve.recplot2.seqdepth <- function
     ){
   if(!inherits(x, "enve.RecPlot2"))
     stop("'x' must inherit from class `enve.RecPlot2`")
-  pos.cnts.in <- x$pos.counts.in
+  if(low.identity){
+    pos.cnts.in <- x$pos.counts.out
+  }else{
+    pos.cnts.in <- x$pos.counts.in
+  }
   pos.breaks  <- x$pos.breaks
   pos.binsize <- (pos.breaks[-1] - pos.breaks[-length(pos.breaks)])
   seqdepth.in <- pos.cnts.in/pos.binsize
@@ -987,6 +1047,7 @@ enve.recplot2.findPeaks.__emauto_one <- function
   ### Internal ancilliary function (see `enve.recplot2.findPeaks.emauto).
     (x, comp, do_crit, best, verbose, ...){
   peaks <- enve.recplot2.findPeaks.em(x=x, components=comp, ...)
+  if(length(peaks)==0) return(best)
   k <- comp*3 - 1 # mean & sd for each component, and n-1 free alpha parameters
   crit <- do_crit(peaks[[1]]$err.res, k, peaks[[1]]$n.total)
   if(verbose) cat(comp,'\t| LL =', peaks[[1]]$err.res, '\t| Estimate =', crit,
@@ -1049,7 +1110,7 @@ enve.recplot2.findPeaks.__mow_one <- function
    
    # Find peak
    o <- mlv.opts; o$x = lsd1;
-   mode1 <- do.call(mlv, o)$M;
+   mode1 <- median(lsd1); # mode1 <- do.call(mlv, o)$M;
    if(verbose) cat('Anchoring at mode =',mode1,'\n')
    param.hat <- fitdist.opts$start; last.hat <- param.hat;
    lim <- NA;

data/utils/enveomics/enveomics.R/R/utils.R

@@ -7,10 +7,28 @@ enve.col.alpha <- function
     ### such as 'darkred' or '#009988'.
     alpha=1/2
     ### Alpha value to add to the color, from 0 to 1.
-  ){
+    ){
   return(
     apply(col2rgb(col), 2,
       function(x) do.call(rgb, as.list(c(x[1:3]/256, alpha))) ) )
   ### Returns a color or a vector of colors in hex notation including alpha.
 }
 
+enve.truncate <- function
+  ### Removes the `n` highest and lowest values from a vector, and applies a
+  ### summary function. The value of `n` is determined such that the central
+  ### range is used, corresponding to the `f` fraction of values.
+    (x,
+    ### A vector of numbers.
+    f=0.95,
+    ### The fraction of values to retain.
+    FUN=mean
+    ### Summary function to apply to the vectors. To obtain the truncated
+    ### vector itself, use `c`.
+    ){
+  n <- round(length(x)*(1-f)/2)
+  y <- sort(x)[ -c(seq(1, n), seq(length(x)+1-n, length(x))) ]
+  return(FUN(y))
+  ### Returns the summary (`FUN`) of the truncated vector.
+}
+

data/utils/enveomics/enveomics.R/README.md

@@ -32,6 +32,7 @@ And open help messages using any of the following commands:
 ?enve.recplot2.changeCutoff
 ?enve.recplot2.findPeaks
 ?enve.recplot2.corePeak
+?enve.recplot2.windowDepthThreshold
 ?enve.recplot2.extractWindows
 ?enve.recplot2.coordinates
 ?enve.recplot2.seqdepth
@@ -41,6 +42,7 @@ And open help messages using any of the following commands:
 ?enve.tribs.test
 ?enve.growthcurve
 ?enve.col.alpha
+?enve.truncate
 ```
 
 You can run some examples using these libraries in the
@@ -50,6 +52,15 @@ For additional information on recruitment plots, see the
 [Recruitment plots working document](https://github.com/lmrodriguezr/enveomics/blob/master/Docs/recplot2.md).
 
 ## Changelog
+* 1.4.4: Removes modeest library as requirement, and replaces mower peak-finder
+  initialization to median (instead of mode).
+* 1.4.2: Solved bug #36.
+* 1.4.0: New option `pos.breaks.tsv` for `enve.recplot2`.
+* 1.3.4: Gracefully handles and plots recruitment plots with insufficient data
+  to find peaks.
+* 1.3.3: New function `enve.recplot2.windowDepthThreshold`.
+* 1.3.2: New option `panel.fun` for `plot.enve.RecPlot2`.
+* 1.3.1: New function enve.truncate.
 * 1.3: Several bug fixes and new utilities for recruitment plots (recplot2).
 * 1.1.0: New function enve.growthcurve and related class enve.GrowthCurve
   with S3 methods plot and summary.

data/utils/enveomics/enveomics.R/man/enve.recplot.Rd

@@ -37,7 +37,7 @@ If non-zero, requires the stats package.}
   \item{main}{Title of the plot.}
   \item{contig.col}{Color of the Contig boundaries. Set to NA to ignore Contig boundaries.}
   \item{ret.recplot}{Indicates if the matrix of the recruitment plot is to be returned.}
-  \item{ret.hist}{Indicates if the vectors of the identity and position histograms are to be returned.}
+  \item{ret.hist}{Ignored, for backwards compatibility.}
   \item{ret.mode}{Indicates if the mode of the identity is to be computed. It requires the modeest
 package.}
   \item{id.cutoff}{Minimum identity to consider an alignment as "top". By default, it is 0.95 for the
@@ -59,7 +59,7 @@ id.mean: Mean identity.
 
 id.median: Median identity.
 
-id.mode (if ret.mode=TRUE): Mode of the identity.
+id.mode (if ret.mode=TRUE): Mode of the identity. Deprecated.
 
 id.hist (if ret.hist=TRUE): Values of the identity histogram.
 

data/utils/enveomics/enveomics.R/man/enve.recplot2-class.Rd

@@ -17,6 +17,7 @@ be produced by `enve.recplot2` and supports S4 method plot.}
     \item{\code{id.counts}:}{(\code{numeric}) Counts per ID bin.}
     \item{\code{id.breaks}:}{(\code{numeric}) Breaks of identity bins.}
     \item{\code{pos.breaks}:}{(\code{numeric}) Breaks of position bins.}
+    \item{\code{pos.names}:}{(\code{character}) Names of the position bins.}
     \item{\code{seq.breaks}:}{(\code{numeric}) Breaks of input sequences.}
     \item{\code{peaks}:}{(\code{list}) Peaks identified in the recplot.
 Limits of the subject sequences after concatenation.}

data/utils/enveomics/enveomics.R/man/enve.recplot2.Rd

@@ -3,10 +3,10 @@
 \title{enve recplot2}
 \description{Produces recruitment plots provided that BlastTab.catsbj.pl has
 been previously executed.}
-\usage{enve.recplot2(prefix, plot = TRUE, pos.breaks = 1000, id.breaks = 300, 
-    id.free.range = FALSE, id.metric = c("identity", "corrected identity", 
-        "bit score"), id.summary = sum, id.cutoff = 95, threads = 2, 
-    verbose = TRUE, ...)}
+\usage{enve.recplot2(prefix, plot = TRUE, pos.breaks = 1000, pos.breaks.tsv = NA, 
+    id.breaks = 300, id.free.range = FALSE, id.metric = c("identity", 
+        "corrected identity", "bit score"), id.summary = sum, 
+    id.cutoff = 95, threads = 2, verbose = TRUE, ...)}
 \arguments{
   \item{prefix}{Path to the prefix of the BlastTab.catsbj.pl output files. At
 least the files .rec and .lim must exist with this prefix.}
@@ -14,7 +14,15 @@ least the files .rec and .lim must exist with this prefix.}
   \item{pos.breaks}{Breaks in the positions histogram. It can also be a vector of break
 points, and values outside the range are ignored. If zero (0), it
 uses the sequence breaks as defined in the .lim file, which means
-one bin per contig (or gene, if the mapping is agains genes).}
+one bin per contig (or gene, if the mapping is agains genes). Ignored
+if `pos.breaks.tsv` is passed.}
+  \item{pos.breaks.tsv}{Path to a list of (absolute) coordinates to use as position breaks.
+This tab-delimited file can be produced by `GFF.catsbj.pl`, and it
+must contain at least one column: coordinates of the break positions of
+each position bin. If it has a second column, this is used as the name
+of the position bin that ends at the given coordinate (the first row is
+ignored). Any additional columns are currently ignored. If NA,
+position bins are determined by `pos.breaks`.}
   \item{id.breaks}{Breaks in the identity histogram. It can also be a vector of break
 points, and values outside the range are ignored.}
   \item{id.free.range}{Indicates that the range should be freely set from the observed