fixed editCstring & editClist to handle matrix input

edited README.md

R/moses2.R

@@ -111,7 +111,7 @@ parseMout <- function(mout, strip = FALSE) {
 # 
 # Balanced Accuracy = (Sensitivity+Specificity)/2
 
-# translate n >=2 argument boolean operators
+# define n >=2 argument boolean operators
 and <- function(x) Reduce("&", x)
 or <- function(x) Reduce("|", x)
 
@@ -138,11 +138,13 @@ cml2df <- function(cmlist) {
  return(as.data.frame(out))
 }
 
-# evaluate combos from output of moses2combo() on test dfs and return result lists using caret::confusionMatrix
+# evaluate combos from output of moses2combo() on their corresponding test df and return result lists using caret::confusionMatrix
+# return.cm is flag for return of confusion matrix output
 testCstring <- function(combos, testdf, casecol = 1, caserat, return.cm = TRUE) {
+ if(class(testdf) != "data.frame") testdf <- as.data.frame(testdf)
  case <- testdf[[casecol]]
  n <- length(case)
- attach(testdf)
+ attach(testdf, warn = FALSE)
  m <- length(combos)
  results <- matrix(nrow = m,ncol = n, dimnames = list(combos, row.names(testdf)))
  if(return.cm) {
@@ -159,6 +161,8 @@ testCstring <- function(combos, testdf, casecol = 1, caserat, return.cm = TRUE)
  results <- as.data.frame(rbind(case, results))
  return(list(result = results, score = metrics))
  }
+
+ # if not returning confusion matrix do 
  for(i in 1:m){
  results[i,] <- as.numeric(evalstring(combo.edit(combos[i])))
  }
@@ -168,13 +172,11 @@ testCstring <- function(combos, testdf, casecol = 1, caserat, return.cm = TRUE)
 }
 
 # evaluate list of combo strings & compute catagorization metrics. "cc_ratio" is cases/1's over cases + controls/0's
+# returns list of listed pairs of score & confusion matrices merged together seperately for training sets and test sets
 testClist <- function(clist, tdatlist, caseCol = 1) {
 
- # check data list
- for(i in 1:length(tdatlist)) if(!is.data.frame(tdatlist[[i]])) tdatlist[[i]] <- lapply(tdatlist, as.data.frame)
-
  # compute case to control ratio
- case <- tdatlist$test[[1]][[caseCol]]
+ case <- tdatlist$test[[1]][, caseCol]
  cc_ratio <- sum(case) / length(case)
  if(is.nan(cc_ratio) | cc_ratio == 0) stop("malformed case column")
 

README.md

@@ -49,13 +49,13 @@ system2() help page for other values and other system2() variable options.
 **run combos on training and testing sets and generate confusion matrix** 
 `scoresNconfusionMatrix <- testClist(combosNscores$combo, listOfDataPartitions, fraction_of_cases_in_sample)` 
 
-**aggregate fold results with scores and confusion matrix** 
+**combine fold results and generate aggregate score and confusion matrix** 
 `aggScores <- aggResults(scoresNconfusionMatrix)` 
 
 *the "aggScores" dataframe is ranked by score so combos can be filtered by row index* 
 
 **make dataframe of genes/feataures** 
-`featureCount <- combo2fcount(names(aggScores))` 
+`featureCount <- combo2fcount(names(aggScores[[2]]))` 
 
 **TODO:** 
 establish and implement a score cutoff to filter the combos