@@ -1,8 +1,8 @@

 Package: MotifDb

 Type: Package

 Title: An Annotated Collection of Protein-DNA Binding Sequence Motifs

-Version: 1.19.22

-Date: 2018-01-23

+Version: 1.21.2

+Date: 2018-01-31

 Author: Paul Shannon, Matt Richards

 Maintainer: Paul Shannon <pshannon@systemsbiology.org>

 Depends: R (>= 2.15.0), methods, BiocGenerics, S4Vectors, IRanges, Biostrings

@@ -1,7 +1,7 @@

 setGeneric('query', signature='object', function(object, queryString, ignore.case=TRUE)

               standardGeneric ('query'))

 setGeneric('motifToGene', signature='object', function(object, motifs, source) standardGeneric('motifToGene'))

-setGeneric('geneToMotif', signature='object', function(object, geneSymbols, source) standardGeneric('geneToMotif'))

+setGeneric('geneToMotif', signature='object', function(object, geneSymbols, source, ignore.case=FALSE) standardGeneric('geneToMotif'))

 setGeneric('associateTranscriptionFactors', signature='object',

            function(object, tbl.withMotifs,  source, expand.rows) standardGeneric('associateTranscriptionFactors'))

 #------------------------------------------------------------------------------------------------------------------------

@@ -359,7 +359,8 @@ setMethod ('motifToGene', 'MotifList',

         tbl <- unique(tbl [, c("geneSymbol", "providerId", "dataSource", "organism", "pubmedID")])

         colnames(tbl) <- c("geneSymbol", "motif", "dataSource", "organism", "pubmedID")

         tbl <- tbl[, c("motif", "geneSymbol", "dataSource", "organism", "pubmedID")]

-        tbl$source <- "MotifDb"

+        if(nrow(tbl) > 0)

+           tbl$source <- "MotifDb"

         }

      if(source %in% c("tfclass")){

         motif <- NULL

@@ -370,7 +371,8 @@ setMethod ('motifToGene', 'MotifList',

         tbl <- tbl[order(tbl$motif),]

         rownames(tbl) <- NULL

         colnames(tbl) <- c("motif", "geneSymbol", "pubmedID")

-        tbl$source <- "TFClass"

+        if(nrow(tbl) > 0)

+           tbl$source <- "TFClass"

         }

      tbl

      })

@@ -379,12 +381,16 @@ setMethod ('motifToGene', 'MotifList',

 # returns a data.frame with motif, geneSymbol, source, pubmedID columns

 setMethod ('geneToMotif', 'MotifList',

-   function (object, geneSymbols, source) {

+   function (object, geneSymbols, source, ignore.case=FALSE) {

      source <- tolower(source)

      stopifnot(source %in% c("motifdb", "tfclass"))

      extract.mdb <- function(gene){

         geneSymbol <- NULL # workaround the R CMD check "no visible binding for global variable"

-        tbl <- as.data.frame(subset(mcols(object), geneSymbol == gene))

+        if(ignore.case)

+           tbl <- as.data.frame(subset(mcols(object), tolower(geneSymbol) == tolower(gene)))

+        else

+           tbl <- as.data.frame(subset(mcols(object), geneSymbol == gene))

+

         tbl <- unique(tbl [, c("geneSymbol", "providerId", "dataSource", "organism", "pubmedID")])

         colnames(tbl) <- c("geneSymbol", "motif", "dataSource", "organism", "pubmedID")

         tbl

@@ -392,17 +398,22 @@ setMethod ('geneToMotif', 'MotifList',

      if(source %in% c("motifdb")){

         tbls <- lapply(geneSymbols, extract.mdb)

         result <- do.call(rbind, tbls)

-        result$source <- "MotifDb"

+        if(nrow(result) > 0)

+           result$source <- "MotifDb"

         }

      if(source %in% c("tfclass")){

-        tbl <- subset(object@manuallyCuratedGeneMotifAssociationTable, tf.gene %in% geneSymbols)

+        if(ignore.case)

+           tbl <- subset(object@manuallyCuratedGeneMotifAssociationTable, tolower(tf.gene) %in% tolower(geneSymbols))

+        else

+           tbl <- subset(object@manuallyCuratedGeneMotifAssociationTable, tf.gene %in% geneSymbols)

         tf.gene <- NULL; motif <- NULL  # workaround R CMD CHECK "no visible binding ..." bogus error

         tbl <- unique(tbl[, c("motif", "tf.gene", "pubmedID")])

         tbl <- tbl[order(tbl$tf.gene),]

         rownames(tbl) <- NULL

         colnames(tbl) <- c("motif", "geneSymbol", "pubmedID")

         result <- tbl[, c("geneSymbol", "motif", "pubmedID")]

-        result$source <- "TFClass"

+        if(nrow(result) > 0)

+           result$source <- "TFClass"

         }

      result

      })

@@ -43,6 +43,7 @@ runTests = function ()

   test.geneToMotif()

   test.geneToMotif.ignore.jasparSuffixes()

+  test.geneToMotif.oneGene.noMotifs

   test.motifToGene()

   test.associateTranscriptionFactors()

@@ -762,13 +763,15 @@ test.geneToMotif <- function()

    printf("--- test.geneToMotif")

    mdb <- MotifDb

-   genes <- c("FOS", "ATF5", "bogus")

-

+   genes <- c("FOS", "ATF5", "bogus", "SATB2")

+   good.genes <- genes[-which(genes=="bogus")]

       # use  TFClass family classifcation

    tbl.tfClass <- geneToMotif(mdb, genes, source="TfClaSS")   # intentional mis-capitalization

-   checkEquals(sort(tbl.tfClass$gene),  sort(c("ATF5", "FOS", "FOS")))

-   checkEquals(sort(tbl.tfClass$motif),  sort(c("MA0833.1", "MA0099.2", "MA0476.1")))

-   checkEquals(tbl.tfClass$source, rep("TFClass", 3))

+   checkTrue(all(good.genes %in% tbl.tfClass$gene))

+

+   expected.motifs <- c("MA0833.1", "MA0099.2", "MA0476.1", "MA0679.1", "MA0754.1", "MA0755.1", "MA0756.1", "MA0757.1")

+   checkTrue(all(expected.motifs %in% tbl.tfClass$motif))

+   checkEquals(unique(tbl.tfClass$source), "TFClass")

       # MotifDb mode uses the MotifDb metadata, pulled from many sources

    tbl.mdb <- geneToMotif(mdb, genes, source="mOtifdb")     # intentional mis-capitalization

@@ -778,8 +781,30 @@ test.geneToMotif <- function()

       # MotifDb for ATF5

       # todo: compare the MA0110596_1.02 matrix of cisp_1.02 to japar MA0833.1

+     # check use of ignore.case

+   tbl.caseSensitive <-  geneToMotif(MotifDb, "STAT4", source="MotifDb")

+   checkEquals(length(grep("jaspar", tbl.caseSensitive$dataSource, ignore.case=TRUE)), 0)

+   tbl.caseInsensitive <-  geneToMotif(MotifDb, "STAT4", source="MotifDb", ignore.case=TRUE)

+   checkTrue(length(grep("jaspar", tbl.caseInsensitive$dataSource, ignore.case=TRUE)) >= 3)

+

+   tbl.caseSensitive <-  geneToMotif(MotifDb, "stat4", source="TFclass")

+   checkEquals(nrow(tbl.caseSensitive), 0)

+   tbl.caseInsensitive <-  geneToMotif(MotifDb, "stat4", source="TFclass", ignore.case=TRUE)

+   checkTrue(nrow(tbl.caseInsensitive) >= 5)

+

 } # test.geneToMotif

 #------------------------------------------------------------------------------------------------------------------------

+# this case discovered (31 jan 2018). when called on a gene/source combination for which there are

+# no motifs, i attempted to add the mapping source (either "MotifDb", "TFClass") as a column

+# to an empty data.frame.  check for that and its fix here

+test.geneToMotif.oneGene.noMotifs <- function()

+{

+   checkEquals(nrow(geneToMotif(MotifDb, "SATB2", "MotifDb")), 0)

+   checkEquals(nrow(geneToMotif(MotifDb, "bogus-arandum", "MotifDb")), 0)

+   checkEquals(nrow(geneToMotif(MotifDb, "bogus-arandum", "TFclass")), 0)

+

+} # test.geneToMotif.oneGene.noMotifs

+#------------------------------------------------------------------------------------------------------------------------

 # sad to say I do not recall what problem/fix is tested here (pshannon, 23 jan 2018).

 # however, it demonstrates the variety of results which can be returned by non-jaspar datasets

 # when using the MotifDb mapping source, and the relative paucity which is sometimes

@@ -14,12 +14,13 @@ The MotifDb source is in fact the usually 1:1 gene/motif mapping

 provided by each of the data sources upon which MotifDb is built.

 }

 \usage{

-\S4method{geneToMotif}{MotifList}(object, geneSymbols, source)

+\S4method{geneToMotif}{MotifList}(object, geneSymbols, source, ignore.case)

 }

 \arguments{

   \item{object}{a \code{MotifList} object.}

   \item{geneSymbols}{a \code{character} string}

-  \item{source}{a \code{character} string, either 'MotifDb' or "TFclass' (case insensitive)}

+  \item{source}{a \code{character} string, either 'MotifDb' or 'TFclass' (case insensitive)}

+  \item{ignore.case}{a \code{logical} variable, default FALSE, guiding gene name matching}

 }

 \value{