@@ -8,8 +8,6 @@

 ### list.

 ###

-setClassUnion("integer_OR_NULL", c("integer", "NULL"))

-

 setClass("GsnapParam",

          representation(max_mismatches = "integer_OR_NULL",

                         suboptimal_levels = "integer",

@@ -8,13 +8,13 @@

 ### list.

 ###

-setClassUnion("integerORNULL", c("integer", "NULL"))

+setClassUnion("integer_OR_NULL", c("integer", "NULL"))

 setClass("GsnapParam",

-         representation(max_mismatches = "integerORNULL",

+         representation(max_mismatches = "integer_OR_NULL",

                         suboptimal_levels = "integer",

                         novelsplicing = "logical",

-                        splicing = "characterORNULL",

+                        splicing = "character_OR_NULL",

                         terminal_threshold = "integer",

                         gmap_mode = "character",

                         clip_overlap = "logical"),

@@ -205,12 +205,12 @@ setAs("GsnapParam", "list", function(from) {

 as.list.GmapAlignerParam <- function(x, ...) as(x, "list")

-setAs("ANY", "characterORNULL", function(from) {

+setAs("ANY", "character_OR_NULL", function(from) {

   if (is.null(from))

     NULL

   else as.character(from)

 })

-setAs("ANY", "integerORNULL", function(from) {

+setAs("ANY", "integer_OR_NULL", function(from) {

   if (is.null(from))

     NULL

   else as.integer(from)

@@ -203,7 +203,7 @@ setAs("GsnapParam", "list", function(from) {

           to

       })

-as.list.GmapAlignerParam <- function(x) as(x, "list")

+as.list.GmapAlignerParam <- function(x, ...) as(x, "list")

 setAs("ANY", "characterORNULL", function(from) {

   if (is.null(from))

@@ -9,27 +9,16 @@

 ###

 setClassUnion("integerORNULL", c("integer", "NULL"))

-setClassUnion("GmapSnpsORNULL", c("GmapSnps", "NULL"))

 setClass("GsnapParam",

-         representation(genome = "GmapGenome",

-                        part = "characterORNULL", # used by parallelized_gsnap

-                        batch = "character", # weird "0", "1", ... 

-                        max_mismatches = "integerORNULL",

+         representation(max_mismatches = "integerORNULL",

                         suboptimal_levels = "integer",

-                        snps = "GmapSnpsORNULL",

-                        mode = "character",

-                        nthreads = "integer",

                         novelsplicing = "logical",

                         splicing = "characterORNULL",

-                        npaths = "integer",

-                        quiet_if_excessive = "logical",

-                        nofails = "logical", 

-                        split_output = "logical",

                         terminal_threshold = "integer",

                         gmap_mode = "character",

-                        clip_overlap = "logical",

-                        extra = "list"))

+                        clip_overlap = "logical"),

+         contains="GmapAlignerParam")

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

 ### Accessors

@@ -148,6 +137,27 @@ gsnap_extra <- function(x) {

 ### Constructor

 ###

+newGmapAlignerParam <- function(Class, genome, snps) {

+    if (missing(genome))

+        stop("The 'genome' must be specified (and coercible to GmapGenome)")

+    args <- formals(sys.function(sys.parent(1L)))

+    params <- mget(names(args), parent.frame())

+    params$unique_only <- NULL

+    paramClasses <- getSlots(Class)

+    paramClasses <- paramClasses[setdiff(names(paramClasses),

+                                         c("extra", "snps"))]

+    params <- mapply(as, params[names(paramClasses)], paramClasses,

+                     SIMPLIFY = FALSE)

+    if (!is.null(snps)) {

+        if (!is(snps, "GmapSnps")) {

+            snps <- GmapSnps(snps, genome)

+        }

+        params$snps <- snps

+    }

+    params$extra <- evalq(list(...), parent.frame())

+    do.call(new, c(Class, params))

+}

+

 GsnapParam <- function(genome, unique_only = FALSE, molecule = c("RNA", "DNA"),

                        max_mismatches = NULL,

                        suboptimal_levels = 0L, mode = "standard",

@@ -157,54 +167,43 @@ GsnapParam <- function(genome, unique_only = FALSE, molecule = c("RNA", "DNA"),

                        split_output = !unique_only,

                        novelsplicing = FALSE, splicing = NULL, 

                        nthreads = 1L, part = NULL, batch = "2",

-                       terminal_threshold = if (molecule == "DNA") 1000L else 2L,

-                       gmap_mode = if (molecule == "DNA") "none" else

-                       "pairsearch,terminal,improve",

+                       terminal_threshold =

+                           if (molecule == "DNA") 1000L else 2L,

+                       gmap_mode = if (molecule == "DNA") "none"

+                                   else "pairsearch,terminal,improve",

                        clip_overlap = FALSE, ...)

 {

-  if (missing(genome))

-    stop("The 'genome' must be specified (should be coercible to GmapGenome)")

   molecule <- match.arg(molecule)

-  args <- formals(sys.function())

-  params <- mget(names(args), environment())

-  params$unique_only <- NULL

-  paramClasses <- getSlots("GsnapParam")

-  paramClasses <- paramClasses[setdiff(names(paramClasses), c("extra", "snps"))]

-  params <- mapply(as, params[names(paramClasses)], paramClasses,

-                   SIMPLIFY = FALSE)

-  if (!is.null(snps)) {

-    if (!is(snps, "GmapSnps")) {

-      snps <- GmapSnps(snps, genome)

-    }

-    params$snps <- snps

-  }

-  params$extra <- list(...)

-  do.call(new, c("GsnapParam", params))

+  newGmapAlignerParam("GsnapParam", genome, snps)

 }

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

 ### Coercion

 ###

-setAs("GsnapParam", "list", function(from) {

+GmapAlignerParam_asList <- function(from) {

   to <- lapply(slotNames(from), slot, object = from)

   names(to) <- slotNames(from)

-  to$split_output <- if (to$split_output) "gsnap" else NULL

+  to$split_output <- if (to$split_output) tolower(sub("Param", "", class(from)))

   to$db <- genome(to$genome)

   to$dir <- path(directory(to$genome))

   to$genome <- NULL

   to$use_snps <- name(to$snps)

   to$snpsdir <- path(directory(to$snps))

-  to$novelsplicing <- as.integer(to$novelsplicing)

-  to <- rename(to, splicing = "use_splicing")

   extras <- to$extra

   to <- c(to, extras)

-  to$extra <- NULL

-  

+  to$extra <- NULL  

   to

-})

+}

+

+setAs("GsnapParam", "list", function(from) {

+          to <- GmapAlignerParam_asList(from)

+          to$novelsplicing <- as.integer(to$novelsplicing)

+          to <- rename(to, splicing = "use_splicing")

+          to

+      })

-setMethod("as.list", "GsnapParam", function(x) as(x, "list"))

+as.list.GmapAlignerParam <- function(x) as(x, "list")

 setAs("ANY", "characterORNULL", function(from) {

   if (is.null(from))

@@ -222,7 +221,7 @@ setAs("ANY", "integerORNULL", function(from) {

 ### Show

 ###

-setMethod("show", "GsnapParam", function(object) {

+setMethod("show", "GmapAlignerParam", function(object) {

   slots <- lapply(slotNames(object), slot, object = object)

   names(slots) <- slotNames(object)

   slots$genome <- paste0(slots$genome@name,

@@ -231,6 +230,6 @@ setMethod("show", "GsnapParam", function(object) {

     slots$snps <- paste0(name(slots$snps),

                          " (", path(directory(slots$snps)), ")")

   }

-  cat("A GsnapParams object\n",

+  cat("A ", class(object), " object\n",

       paste0(names(slots), ": ", slots, collapse = "\n"), "\n", sep = "")

 })

@@ -28,7 +28,7 @@ setClass("GsnapParam",

                         split_output = "logical",

                         terminal_threshold = "integer",

                         gmap_mode = "character",

-                        clip_overlaps = "logical",

+                        clip_overlap = "logical",

                         extra = "list"))

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

@@ -160,7 +160,7 @@ GsnapParam <- function(genome, unique_only = FALSE, molecule = c("RNA", "DNA"),

                        terminal_threshold = if (molecule == "DNA") 1000L else 2L,

                        gmap_mode = if (molecule == "DNA") "none" else

                        "pairsearch,terminal,improve",

-                       clip_overlaps = FALSE, ...)

+                       clip_overlap = FALSE, ...)

 {

   if (missing(genome))

     stop("The 'genome' must be specified (should be coercible to GmapGenome)")

@@ -28,6 +28,7 @@ setClass("GsnapParam",

                         split_output = "logical",

                         terminal_threshold = "integer",

                         gmap_mode = "character",

+                        clip_overlaps = "logical",

                         extra = "list"))

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

@@ -158,7 +159,8 @@ GsnapParam <- function(genome, unique_only = FALSE, molecule = c("RNA", "DNA"),

                        nthreads = 1L, part = NULL, batch = "2",

                        terminal_threshold = if (molecule == "DNA") 1000L else 2L,

                        gmap_mode = if (molecule == "DNA") "none" else

-                       "pairsearch,terminal,improve", ...)

+                       "pairsearch,terminal,improve",

+                       clip_overlaps = FALSE, ...)

 {

   if (missing(genome))

     stop("The 'genome' must be specified (should be coercible to GmapGenome)")

@@ -27,7 +27,7 @@ setClass("GsnapParam",

                         nofails = "logical", 

                         split_output = "logical",

                         terminal_threshold = "integer",

-                        gmap_mode = "characterORNULL",

+                        gmap_mode = "character",

                         extra = "list"))

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

@@ -157,7 +157,9 @@ GsnapParam <- function(genome, unique_only = FALSE, molecule = c("RNA", "DNA"),

                        novelsplicing = FALSE, splicing = NULL, 

                        nthreads = 1L, part = NULL, batch = "2",

                        terminal_threshold = if (molecule == "DNA") 1000L else 2L,

-                       gmap_mode = if (molecule == "DNA") "none", ...) {

+                       gmap_mode = if (molecule == "DNA") "none" else

+                       "pairsearch,terminal,improve", ...)

+{

   if (missing(genome))

     stop("The 'genome' must be specified (should be coercible to GmapGenome)")

   molecule <- match.arg(molecule)

@@ -27,7 +27,7 @@ setClass("GsnapParam",

                         nofails = "logical", 

                         split_output = "logical",

                         terminal_threshold = "integer",

-                        gmap_mode = "character",

+                        gmap_mode = "characterORNULL",

                         extra = "list"))

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

@@ -26,6 +26,8 @@ setClass("GsnapParam",

                         quiet_if_excessive = "logical",

                         nofails = "logical", 

                         split_output = "logical",

+                        terminal_threshold = "integer",

+                        gmap_mode = "character",

                         extra = "list"))

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

@@ -145,7 +147,7 @@ gsnap_extra <- function(x) {

 ### Constructor

 ###

-GsnapParam <- function(genome, unique_only = FALSE,

+GsnapParam <- function(genome, unique_only = FALSE, molecule = c("RNA", "DNA"),

                        max_mismatches = NULL,

                        suboptimal_levels = 0L, mode = "standard",

                        snps = NULL,

@@ -153,9 +155,12 @@ GsnapParam <- function(genome, unique_only = FALSE,

                        quiet_if_excessive = unique_only, nofails = unique_only,

                        split_output = !unique_only,

                        novelsplicing = FALSE, splicing = NULL, 

-                       nthreads = 1L, part = NULL, batch = "2", ...) {

+                       nthreads = 1L, part = NULL, batch = "2",

+                       terminal_threshold = if (molecule == "DNA") 1000L else 2L,

+                       gmap_mode = if (molecule == "DNA") "none", ...) {

   if (missing(genome))

     stop("The 'genome' must be specified (should be coercible to GmapGenome)")

+  molecule <- match.arg(molecule)

   args <- formals(sys.function())

   params <- mget(names(args), environment())

   params$unique_only <- NULL

@@ -154,6 +154,8 @@ GsnapParam <- function(genome, unique_only = FALSE,

                        split_output = !unique_only,

                        novelsplicing = FALSE, splicing = NULL, 

                        nthreads = 1L, part = NULL, batch = "2", ...) {

+  if (missing(genome))

+    stop("The 'genome' must be specified (should be coercible to GmapGenome)")

   args <- formals(sys.function())

   params <- mget(names(args), environment())

   params$unique_only <- NULL

@@ -182,15 +182,10 @@ setAs("GsnapParam", "list", function(from) {

   to$db <- genome(to$genome)

   to$dir <- path(directory(to$genome))

   to$genome <- NULL

-  if(!is.null(to$snps)) {

-    to$use_snps <- name(to$snps)

-  } else {

-    to$use_snps <- NULL

-  }

+  to$use_snps <- name(to$snps)

   to$snpsdir <- path(directory(to$snps))

-  to$snps <- NULL

   to$novelsplicing <- as.integer(to$novelsplicing)

-

+  to <- rename(to, splicing = "use_splicing")

   extras <- to$extra

   to <- c(to, extras)

   to$extra <- NULL

@@ -189,6 +189,7 @@ setAs("GsnapParam", "list", function(from) {

   }

   to$snpsdir <- path(directory(to$snps))

   to$snps <- NULL

+  to$novelsplicing <- as.integer(to$novelsplicing)

   extras <- to$extra

   to <- c(to, extras)

@@ -154,21 +154,20 @@ GsnapParam <- function(genome, unique_only = FALSE,

                        split_output = !unique_only,

                        novelsplicing = FALSE, splicing = NULL, 

                        nthreads = 1L, part = NULL, batch = "2", ...) {

-  params <- formals(sys.function())

-  mc <- as.list(match.call(expand.dots = FALSE))[-1L]

-  params[names(mc)] <- mc

+  args <- formals(sys.function())

+  params <- mget(names(args), environment())

   params$unique_only <- NULL

+  paramClasses <- getSlots("GsnapParam")

+  paramClasses <- paramClasses[setdiff(names(paramClasses), c("extra", "snps"))]

+  params <- mapply(as, params[names(paramClasses)], paramClasses,

+                   SIMPLIFY = FALSE)

   if (!is.null(snps)) {

     if (!is(snps, "GmapSnps")) {

-      params$snps <- GmapSnps(snps, genome)

+      snps <- GmapSnps(snps, genome)

     }

+    params$snps <- snps

   }

-  dots <- list(...)

-  ##TODO: this breaks if ... has anything in it

-  if (length(dots) > 0) {

-    params$extra <- params$...

-  }

-  params$... <- NULL

+  params$extra <- list(...)

   do.call(new, c("GsnapParam", params))

 }

@@ -200,6 +199,18 @@ setAs("GsnapParam", "list", function(from) {

 setMethod("as.list", "GsnapParam", function(x) as(x, "list"))

+setAs("ANY", "characterORNULL", function(from) {

+  if (is.null(from))

+    NULL

+  else as.character(from)

+})

+setAs("ANY", "integerORNULL", function(from) {

+  if (is.null(from))

+    NULL

+  else as.integer(from)

+})

+

+

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

 ### Show

 ###

@@ -163,9 +163,9 @@ GsnapParam <- function(genome, unique_only = FALSE,

       params$snps <- GmapSnps(snps, genome)

     }

   }

-  extra <- params$...

+  dots <- list(...)

   ##TODO: this breaks if ... has anything in it

-  if (!missing(extra) && !is.null(names(extra))) {

+  if (length(dots) > 0) {

     params$extra <- params$...

   }

   params$... <- NULL

@@ -190,6 +190,11 @@ setAs("GsnapParam", "list", function(from) {

   }

   to$snpsdir <- path(directory(to$snps))

   to$snps <- NULL

+

+  extras <- to$extra

+  to <- c(to, extras)

+  to$extra <- NULL

+  

   to

 })

@@ -152,7 +152,7 @@ GsnapParam <- function(genome, unique_only = FALSE,

                        npaths = if (unique_only) 1L else 100L,

                        quiet_if_excessive = unique_only, nofails = unique_only,

                        split_output = !unique_only,

-                       novelsplicing = 0L, splicing = NULL, 

+                       novelsplicing = FALSE, splicing = NULL, 

                        nthreads = 1L, part = NULL, batch = "2", ...) {

   params <- formals(sys.function())

   mc <- as.list(match.call(expand.dots = FALSE))[-1L]

new file mode 100644

@@ -0,0 +1,213 @@

+### =========================================================================

+### GsnapParam class

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

+###

+### High-level interface to gsnap. As a complex operation, we need a

+### formal parameter object. It should only formally represent the

+### most commonly used parameters. The rest fall into the 'extra'

+### list.

+###

+

+setClassUnion("integerORNULL", c("integer", "NULL"))

+setClassUnion("GmapSnpsORNULL", c("GmapSnps", "NULL"))

+

+setClass("GsnapParam",

+         representation(genome = "GmapGenome",

+                        part = "characterORNULL", # used by parallelized_gsnap

+                        batch = "character", # weird "0", "1", ... 

+                        max_mismatches = "integerORNULL",

+                        suboptimal_levels = "integer",

+                        snps = "GmapSnpsORNULL",

+                        mode = "character",

+                        nthreads = "integer",

+                        novelsplicing = "logical",

+                        splicing = "characterORNULL",

+                        npaths = "integer",

+                        quiet_if_excessive = "logical",

+                        nofails = "logical", 

+                        split_output = "logical",

+                        extra = "list"))

+

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

+### Accessors

+###

+

+gsnap_part <- function(x) {

+  x@part

+}

+gsnap_batch <- function(x) {

+  x@batch

+}

+gsnap_max_mismatches <- function(x) {

+  x@max_mismatches

+}

+gsnap_suboptimal_levels <- function(x) {

+  x@suboptimal_levels

+}

+gsnap_use_snps <- function(x) {

+  x@use_snps

+}

+gsnap_snpsdir <- function(x) {

+  x@snpsdir

+}

+gsnap_mode <- function(x) {

+  x@mode

+}

+gsnap_nthreads <- function(x) {

+  x@nthreads

+}

+gsnap_novelsplicing <- function(x) {

+  x@novelsplicing

+}

+gsnap_splicing <- function(x) {

+  x@splicing

+}

+gsnap_npaths <- function(x) {

+  x@npaths

+}

+gsnap_quiet_if_excessive <- function(x) {

+  x@quiet_if_excessive

+}

+gsnap_nofails <- function(x) {

+  x@nofails

+}

+gsnap_split_output <- function(x) {

+  x@split_output

+}

+gsnap_extra <- function(x) {

+  x@extra

+}

+

+`gsnap_part<-` <- function(x, value) {

+  x@part <- value

+  x

+}

+`gsnap_batch<-` <- function(x, value) {

+  x@batch <- value

+  x

+}

+`gsnap_max_mismatches<-` <- function(x, value) {

+  x@max_mismatches <- value

+  x

+}

+`gsnap_suboptimal_levels<-` <- function(x, value) {

+  x@suboptimal_levels <- value

+  x

+}

+`gsnap_use_snps<-` <- function(x, value) {

+  x@use_snps <- value

+  x

+}

+`gsnap_snpsdir<-` <- function(x, value) {

+  x@snpsdir <- value

+  x

+}

+`gsnap_mode<-` <- function(x, value) {

+  x@mode <- value

+  x

+}

+`gsnap_nthreads<-` <- function(x, value) {

+  x@nthreads <- value

+  x

+}

+`gsnap_novelsplicing<-` <- function(x, value) {

+  x@novelsplicing <- value

+  x

+}

+`gsnap_splicing<-` <- function(x, value) {

+  x@splicing <- value

+  x

+}

+`gsnap_npaths<-` <- function(x, value) {

+  x@npaths <- value

+  x

+}

+`gsnap_quiet_if_excessive<-` <- function(x, value) {

+  x@quiet_if_excessive <- value

+  x

+}

+`gsnap_nofails<-` <- function(x, value) {

+  x@nofails <- value

+  x

+}

+`gsnap_split_output<-` <- function(x, value) {

+  x@split_output <- value

+  x

+}

+`gsnap_extra<-` <- function(x, value) {

+  x@extra <- value

+  x

+}

+

+## etc..

+

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

+### Constructor

+###

+

+GsnapParam <- function(genome, unique_only = FALSE,

+                       max_mismatches = NULL,

+                       suboptimal_levels = 0L, mode = "standard",

+                       snps = NULL,

+                       npaths = if (unique_only) 1L else 100L,

+                       quiet_if_excessive = unique_only, nofails = unique_only,

+                       split_output = !unique_only,

+                       novelsplicing = 0L, splicing = NULL, 

+                       nthreads = 1L, part = NULL, batch = "2", ...) {

+  params <- formals(sys.function())

+  mc <- as.list(match.call(expand.dots = FALSE))[-1L]

+  params[names(mc)] <- mc

+  params$unique_only <- NULL

+  if (!is.null(snps)) {

+    if (!is(snps, "GmapSnps")) {

+      params$snps <- GmapSnps(snps, genome)

+    }

+  }

+  extra <- params$...

+  ##TODO: this breaks if ... has anything in it

+  if (!missing(extra) && !is.null(names(extra))) {

+    params$extra <- params$...

+  }

+  params$... <- NULL

+  do.call(new, c("GsnapParam", params))

+}

+

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

+### Coercion

+###

+

+setAs("GsnapParam", "list", function(from) {

+  to <- lapply(slotNames(from), slot, object = from)

+  names(to) <- slotNames(from)

+  to$split_output <- if (to$split_output) "gsnap" else NULL

+  to$db <- genome(to$genome)

+  to$dir <- path(directory(to$genome))

+  to$genome <- NULL

+  if(!is.null(to$snps)) {

+    to$use_snps <- name(to$snps)

+  } else {

+    to$use_snps <- NULL

+  }

+  to$snpsdir <- path(directory(to$snps))

+  to$snps <- NULL

+  to

+})

+

+setMethod("as.list", "GsnapParam", function(x) as(x, "list"))

+

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

+### Show

+###

+

+setMethod("show", "GsnapParam", function(object) {

+  slots <- lapply(slotNames(object), slot, object = object)

+  names(slots) <- slotNames(object)

+  slots$genome <- paste0(slots$genome@name,

+                         " (", path(directory(slots$genome)), ")")

+  if (!is.null(slots$snps)) {

+    slots$snps <- paste0(name(slots$snps),

+                         " (", path(directory(slots$snps)), ")")

+  }

+  cat("A GsnapParams object\n",

+      paste0(names(slots), ": ", slots, collapse = "\n"), "\n", sep = "")

+})