R/plot.R

In groupschoof/PhyloFun: Annotates Proteins with Gene Ontology (GO) Terms

library(xtable)

goAnnotationAbbreviations <- function( protein.accessions, go.type.annos,
  go.type.anno.space, unknown.anno.abbrev='0', go.type.abbrevs=list(
    'biological_process'='B', 'cellular_component'='C',
    'molecular_function'='M' ) ) {
  # Generates abbreviations for Gene Ontology term annotations, so that i.e. in
  # the plotted phylogenetic tree each protein accession can be suffixed with a
  # six character text in which each two characters consist of the abbreviated
  # GO type followed by the annotated indexed unique GO terms. I.e.: 'B2CøM2'.
  #
  # Args:
  #  protein.accessions  : accessions of the proteins whose annotations shall
  #                        be abbreviated.
  #  go.type.annos       : The result of calling goTypeAnnotationMatrices(
  #                        retrieveUniprotAnnotations( homolog.accessions ), … )
  #  go.type.anno.space  : The result of calling goAnnotationSpaceList(
  #                        go.type.annos, … )
  #  unknown.anno.abbrev : The string to be used to abbreviate the annotation
  #                        'unknown'. Default: 'ø'
  #  go.type.abbrevs     : The abbreviations usd for each GO type, i.e. 'B' for
  #                        'biological_process'
  #
  # Returns: A named list which contains a list of abbreviated GO annotations
  # for each protein accession in anno.mtrx.
  #   
  setNames(
    lapply( protein.accessions, function( acc ) {
      setNames(
        lapply( names( go.type.anno.space ), function( go.type ) {
          paste( go.type.abbrevs[[ go.type ]],
            if ( acc %in% colnames( go.type.annos[[ go.type ]] ) ) {
              anno <- go.type.annos[[ go.type ]][[ 'GO', acc ]]
              which( as.logical(
                lapply( go.type.anno.space[[ go.type ]], identical, anno )
              ) )
            } else {
              unknown.anno.abbrev
            },
            sep=''
          )
        }),
        names( go.type.anno.space )
      )
    }),
    protein.accessions
  )
}

goAnnotationSpaceAbbreviations <- function( go.anno.space,
  abbrev.colnames=c( 'Abbreviation', 'Annotated GO terms' ),
  go.type.abbrevs=list( 'biological_process'='B', 'cellular_component'='C',
    'molecular_function'='M' ), go.con=connectToGeneOntology(),
  close.db.connection=TRUE
  ) {
  # Generates a table for each Gene Ontology term type mapping the unique
  # annotations to an index. This is meant to be the caption for a PhyloFun
  # plotted phylogenetic tree using abbreviated GO term annotations obtained by
  # calling goAnnotationAbbreviations(…).
  #
  # Args:
  #  go.anno.space   : A list of uniquely annotated GO terms for each GO term
  #                    type. Generated by function goAnnotationSpaceList(…).
  #  abbrev.colnames : The column names for each of the abbreviation tables.
  #  go.type.abbrevs : The abbreviations used for each GO term type.
  #  go.con          : Database connection to Gene Ontology
  #  close.db.connection : Set to TRUE the Database Connection to the Gene
  #                    Ontology is automatically closed before this function
  #                    returns.
  #
  # Returns: A list of abbreviation tables, one for each GO term type.
  #   
  gas <- go.anno.space[ as.logical( lapply( names( go.anno.space ),
    function(x) { ! is.null( go.anno.space[[x]] ) } ) ) ]
  rslt <- setNames(
    lapply( names( gas ), function( go.type ) {
      gt.anno.space <- go.anno.space[[ go.type ]]
      max.ind <- length( gt.anno.space )
      abbrevs <- paste( go.type.abbrevs[[ go.type ]],
        as.character( 1:max.ind ), sep=''
      )
      gt.annos <- as.character(
        lapply( go.anno.space[[ go.type ]], function( go.terms ) {
          gt.tbl <- goTermsForAccessionWithLevel( go.terms, con=go.con )
          if ( ! is.null( gt.tbl ) ) {
            toString( lapply( go.terms, function( go.term ) {
              paste( go.term, '(', 
                gt.tbl[ gt.tbl[ , 'acc' ] == go.term, 'name' ][[ 1 ]], ')'
              )
            } ) )
          } else 
            toString( go.terms )
        } )
      )
      matrix( c( abbrevs, gt.annos ),
        nrow=max.ind, dimnames=list( c(), abbrev.colnames )
      )
    }),
    names( gas )
  )
  if ( close.db.connection )
    dbDisconnect( go.con )
  rslt
}

plotPhyloFunTree <- function( query.accession, phyl.tree, query.predictions,
  homolog.go.type.annos, plot.path, device.funk=png, width=2000,
  height.per.leaf=13, min.height=480 ) {
  # Plots the phylogenetic tree to a device specified by 'device.funk'. Width
  # and Height are adjusted to tree's size. GO term annotations are abbreviated
  # and each tip is appended with its respective abbreviations, one for each
  # Gene Ontology term type.
  #
  # Args:
  #  query.accession       : The query proteins accession
  #  phyl.tree             : An object of class phylo representing the
  #                          phylogenetic tree
  #  query.predictions     : The PhyloFun GO term predictions
  #  homolog.go.type.annos : The GO annotations passed to PhyloFun in order to
  #                          annotate the Query
  #  plot.path             : The valid file path to the plot to generate
  #  device.funk           : The function used to redirect the plot to, default
  #                          'png'. Could be set to jpeg, pdf…
  #  width                 : The width in pixel
  #  height.per.leaf       : The height of the plot is this ( argument *
  #                          number.of.tree.tips ).
  #  min.height            : If above product is smaller than this minimum
  #                          height, this minimum will be used instead.
  #                          Default: 480px
  #
  # Returns: A list with two entries. 'caption' holds GO term annotation
  # abbreviations as returned by 'goAnnotationAbbreviations(…)' and
  # 'tree.with.abbrevs' is the phylogenetic tree where the tip labels have
  # their abbreviated GO annotations appended.
  #   

  # Generate a merged GO annotation matrix for both the query and its homologs.
  # One matrix for each GO type:
  all.go.type.annos <- mergeQueryPredictionsAndHomologAnnotations(
    query.accession, query.predictions, homolog.go.type.annos )
  go.type.anno.space <- goAnnotationSpaceList( all.go.type.annos,
    unknown.annot=NULL )
  # Abbreviate each proteins GO annotation one two character string for each GO
  # type:
  anno.abbrevs <- as.character( lapply(
    goAnnotationAbbreviations( phyl.tree$tip.label, all.go.type.annos,
      go.type.anno.space ), paste, collapse=',' ) )
  # Append above abbreviations to the proteins' accessions in the phylogenetic
  # tree:
  phyl.tree$tip.label <- paste( phyl.tree$tip.label, anno.abbrevs )

  hgt <- height.per.leaf * length( phyl.tree$tip.label )
  device.funk( file=plot.path, width=width, 
    height= ( if ( hgt < min.height ) min.height else hgt ) )
  plot( phyl.tree, show.node.label=T )
  axisPhylo()
  dev.off()

  # Generate the plot's caption, that is map the abbreviations to the complete
  # annotations:
  list(
    'caption'=goAnnotationSpaceAbbreviations( go.type.anno.space ),
    'tree.with.abbrevs'=phyl.tree
  )
}

htmlReport <- function( rel.path.2.phyl.tree.plot, query.accession,
  anno.abbrevs, path.2.html.report ) {
  brew( text=REPORT.TEMPLATE, output=path.2.html.report )
}

REPORT.TEMPLATE <- 
'
<!DOCTYPE HTML PUBLIC "-//W3C//DTD HTML 4.01 Transitional//EN"
       "http://www.w3.org/TR/html4/loose.dtd">
<html>
  <head>
    <title>PhyloFun report for <%= query.accession %></title>
    <style type="text/css">
      .table {
        min-width: 820px;
        width: 820px;
        border: 1px solid #e6e6e6;
        border-collapse: collapse;
        -moz-border-radius: 4px;
        -webkit-border-radius: 4px;
        border-radius: 4px;
        color: #171717;
        font-size: 9pt;
      }

      .table td, .table th, .tablescroll_head th {
        padding: 0.5em;
        font-size: 9pt;
        border-bottom: 1px solid #e6e6e6;
      }

      .table th, .tablescroll_head th {
        background-color: #e6e6e6;
      }

      .table td {
        border-right: 1px solid #e6e6e6;
      }
    </style>
  </head>
  <body>
    <div id="phylogenetic_tree_div">
      <img id="phylogenetic_tree_img" src="<%= rel.path.2.phyl.tree.plot %>" class="phyl_tree_img" />
      <div id="phyl_tree_caption_div" class="phyl_tree_caption">
        Phylogenetic maximum likelihood tree for <%= query.accession %>. Node
        labels are Shimodaira-Hasegawa local support values. 
      </div>
      <div id="function_annotation_abbrevs_div">
        <% for ( go.type in names( anno.abbrevs ) ) { %>
          <p>
            <h4><%= go.type %></h4>
            <% print( xtable( anno.abbrevs[[ go.type ]] ), "html" ) %>
          </p>
        <% } %>
      </div>
    </div>
  </body>
</html>
'