Adding Anthony's updates from PR 4 - revert was messy

manifest

@@ -1,6 +1,6 @@
 #Tue Nov 24 19:37:11 UTC 2020
 author=Joshua Gould, Broad Institute
-commandLine=<Rscript> <libdir>main.R <libdir> -d<data.column.name> -o<output.filename> -a<GEO.accession> -f<GEO.SOFT.file> -1<ftp.proxy.server> -2<ftp.proxy.username> -3<ftp.proxy.password> -4<http.proxy.server> -5<http.proxy.username> -6<http.proxy.password>
+commandLine=<Rscript> <libdir>main.R <libdir> -n<Drop.NA.Rows> -d<data.column.name> -o<output.filename> -a<GEO.accession> -f<GEO.SOFT.file> -1<ftp.proxy.server> -2<ftp.proxy.username> -3<ftp.proxy.password> -4<http.proxy.server> -5<http.proxy.username> -6<http.proxy.password>
 cpuType=any
 description=Imports data from the Gene Expression Omnibus (GEO)
 fileFormat=gct
@@ -27,62 +27,75 @@ p2_prefix_when_specified=
 p2_TYPE=FILE
 p2_type=java.io.File
 p2_value=
-p3_default_value=VALUE
-p3_description=The name of the column that contains the expression values
-p3_name=data.column.name
-p3_optional=on
+p3_MODE=
+p3_TYPE=TEXT
+p3_default_value=FALSE
+p3_description=Omit rows containing all "NA" values from GSExxx data series.
+p3_fileFormat=
+p3_flag=-n
+p3_name=Drop.NA.Rows
+p3_numValues=1..1
+p3_optional=
+p3_prefix=
 p3_prefix_when_specified=
 p3_type=java.lang.String
-p3_value=
-p4_default_value=<GEO.accession><GEO.SOFT.file_basename>
-p4_description=The name of the output file
-p4_name=output.filename
-p4_optional=
+p3_value=FALSE\=false;TRUE\=true
+p4_default_value=VALUE
+p4_description=The name of the column that contains the expression values
+p4_name=data.column.name
+p4_optional=on
 p4_prefix_when_specified=
 p4_type=java.lang.String
 p4_value=
-p5_default_value=
-p5_description=FTP Proxy server
-p5_name=ftp.proxy.server
-p5_optional=on
+p5_default_value=<GEO.accession><GEO.SOFT.file_basename>
+p5_description=The name of the output file
+p5_name=output.filename
+p5_optional=
 p5_prefix_when_specified=
 p5_type=java.lang.String
 p5_value=
 p6_default_value=
-p6_description=FTP Proxy username
-p6_name=ftp.proxy.username
+p6_description=FTP Proxy server
+p6_name=ftp.proxy.server
 p6_optional=on
 p6_prefix_when_specified=
 p6_type=java.lang.String
 p6_value=
 p7_default_value=
-p7_description=FTP Proxy password
-p7_name=ftp.proxy.password
+p7_description=FTP Proxy username
+p7_name=ftp.proxy.username
 p7_optional=on
 p7_prefix_when_specified=
-p7_type=PASSWORD
+p7_type=java.lang.String
 p7_value=
 p8_default_value=
-p8_description=HTTP Proxy server
-p8_name=http.proxy.server
+p8_description=FTP Proxy password
+p8_name=ftp.proxy.password
 p8_optional=on
 p8_prefix_when_specified=
-p8_type=java.lang.String
+p8_type=PASSWORD
 p8_value=
 p9_default_value=
-p9_description=HTTP Proxy username
-p9_name=http.proxy.username
+p9_description=HTTP Proxy server
+p9_name=http.proxy.server
 p9_optional=on
 p9_prefix_when_specified=
 p9_type=java.lang.String
 p9_value=
 p10_default_value=
-p10_description=HTTP Proxy password
-p10_name=http.proxy.password
+p10_description=HTTP Proxy username
+p10_name=http.proxy.username
 p10_optional=on
 p10_prefix_when_specified=
-p10_type=PASSWORD
+p10_type=java.lang.String
 p10_value=
+p11_default_value=
+p11_description=HTTP Proxy password
+p11_name=http.proxy.password
+p11_optional=on
+p11_prefix_when_specified=
+p11_type=PASSWORD
+p11_value=
 
 privacy=public
 publicationDate=11/01/2007 20\:31 

src/main.R

@@ -4,7 +4,7 @@ message <- function (..., domain = NULL, appendLF = TRUE) {
 
 # GSExxx, Series
 GseToGct <- function(gse=NULL, data.column.name='VALUE',
-					gct.output.filename=NULL) {
+					gct.output.filename=NULL, omit.na=FALSE) {
 	t <- Table(GPLList(gse)[[1]])
 	ids <- t$ID # ids in GPL
 
@@ -24,7 +24,14 @@ GseToGct <- function(gse=NULL, data.column.name='VALUE',
 		mymatch <- match(ids, tab$ID_REF)
 		return(tab[,data.column.name][mymatch])
 	}))
-
+
+	if(omit.na==TRUE) {
+		filter <- which(rowSums(is.na(data.matrix)) != ncol(data.matrix))
+		data.matrix <- data.matrix[filter,]
+		desc <- desc[filter]
+		ids<- ids[filter]
+	}
+
 	row.names(data.matrix) <- ids
 	gct <- list(data=data.matrix, row.descriptions=desc)
 	write.gct(gct, gct.output.filename)
@@ -33,7 +40,8 @@ GseToGct <- function(gse=NULL, data.column.name='VALUE',
 
 
 # GDSxxx, e.g. GDS1, GDS2577
-GdsToGct <- function(gds=NULL, gct.output.filename) {
+GdsToGct <- function(gds=NULL, gct.output.filename,
+					omit.na=FALSE) {
 	eset <- GDS2eSet(gds, do.log2 = FALSE)
 	f <- eset@featureData
 	annotations <- row.names(varMetadata(f))
@@ -111,6 +119,8 @@ run <- function(libdir, args) {
 			geo.id <- value
 		} else if(flag=='-f') {
 			filename <- value
+		} else if(flag=='-n') {
+			omit.na <- as.logical(value)
 		} else if(flag=='-d') {
 			data.column.name <- value
 		} else if(flag=='-o') {
@@ -170,7 +180,7 @@ run <- function(libdir, args) {
 	if (class(geo.query) == "GSE") {
 		info("converting GSE...") 
 		GseToGct(gse=geo.query, data.column.name=data.column.name,
-			gct.output.filename=gct.output.filename)
+			gct.output.filename=gct.output.filename, omit.na=omit.na)
 	} else if (class(geo.query) == "GDS") {
 		info("converting GDS...")
 		GdsToGct(gds=geo.query, gct.output.filename=gct.output.filename)