fix the jaccard_coef bug

DESCRIPTION

@@ -50,7 +50,8 @@ Imports:
     stats,
     biocViews,
     RANN(>= 2.5),
-    Rcpp
+    Rcpp (>= 0.12.0)
+LinkingTo: Rcpp
 Suggests:
     destiny, 
     Hmisc,
@@ -71,4 +72,3 @@ biocViews: Sequencing, RNASeq, GeneExpression, DifferentialExpression,
     Clustering, MultipleComparison, QualityControl
 Packaged: 2014-04-07 21:31:54 UTC; sarora
 RoxygenNote: 6.0.1
-LinkingTo: Rcpp

NEWS

@@ -1,5 +1,12 @@
 Monocle 2.0.0 Series NEWS
 ================================================================================
+Version 2.6.0
+--------------------------------------------------------------------------------
+  o knn-based density peak clustering is not general for all datasets. Rolled back to 
+    the previous densityPeak clustering algorithm and set it to be the default algorithm.
+    A new Louvain clustering algorithm for dealing with large datasets (like > 500 k cells)
+    is added. 
+  o A few bug fixes for importCDS, exportCDS functions.
 
 Version 2.5.0
 --------------------------------------------------------------------------------

R/clustering.R

@@ -74,10 +74,12 @@ clusterGenes<-function(expr_matrix, k, method=function(x){as.dist((1 - cor(Matri
 #' @importFrom igraph graph.data.frame cluster_louvain modularity membership
 #' @import ggplot2
 #' @importFrom RANN nn2
-#' @useDynLib monocle
 #' @references Rodriguez, A., & Laio, A. (2014). Clustering by fast search and find of density peaks. Science, 344(6191), 1492-1496. doi:10.1126/science.1242072
 #' @references Vincent D. Blondel, Jean-Loup Guillaume, Renaud Lambiotte, Etienne Lefebvre: Fast unfolding of communities in large networks. J. Stat. Mech. (2008) P10008
 #' @references Jacob H. Levine and et.al. Data-Driven Phenotypic Dissection of AML Reveals Progenitor-like Cells that Correlate with Prognosis. Cell, 2015. 
+#' 
+#' @useDynLib monocle
+#' 
 #' @export
 
 clusterCells <- function(cds, 

src/RcppExports.cpp

@@ -6,14 +6,14 @@
 using namespace Rcpp;
 
 // jaccard_coeff
-NumericMatrix jaccard_coeff(NumericMatrix idx, bool weight);
-RcppExport SEXP _monocle_jaccard_coeff(SEXP idxSEXP, SEXP weightSEXP) {
+NumericMatrix jaccard_coeff(SEXP R_idx, SEXP R_weight);
+RcppExport SEXP _monocle_jaccard_coeff(SEXP R_idxSEXP, SEXP R_weightSEXP) {
 BEGIN_RCPP
     Rcpp::RObject rcpp_result_gen;
     Rcpp::RNGScope rcpp_rngScope_gen;
-    Rcpp::traits::input_parameter< NumericMatrix >::type idx(idxSEXP);
-    Rcpp::traits::input_parameter< bool >::type weight(weightSEXP);
-    rcpp_result_gen = Rcpp::wrap(jaccard_coeff(idx, weight));
+    Rcpp::traits::input_parameter< SEXP >::type R_idx(R_idxSEXP);
+    Rcpp::traits::input_parameter< SEXP >::type R_weight(R_weightSEXP);
+    rcpp_result_gen = Rcpp::wrap(jaccard_coeff(R_idx, R_weight));
     return rcpp_result_gen;
 END_RCPP
 }

src/clustering.cpp

@@ -6,14 +6,11 @@ using namespace Rcpp;
 // Weights of both i->j and j->i are recorded if they have intersection. In this case
 // w(i->j) should be equal to w(j->i). In some case i->j has weights while j<-i has no
 // intersections, only w(i->j) is recorded. This is determinded in code `if(u>0)`. 
-// In this way, the undirected graph is symmetrized by halfing the weight 
-// in code `weights(r, 2) = u/(2.0*ncol - u)/2`.
+// The original method described in the phenograph paper is used to calculate the weight. 
 //
 // Author: Chen Hao, Date: 25/09/2015; updated by Xiaojie Qiu Nov. 12, 2017
 
-
-// [[Rcpp::export]]
-NumericMatrix jaccard_coeff(NumericMatrix idx, bool weight) {
+NumericMatrix jaccard_coeff_cpp(NumericMatrix idx, bool weight) {
   int nrow = idx.nrow(), ncol = idx.ncol(), r = 0;
   NumericMatrix weights(nrow*ncol, 3);
 
@@ -52,6 +49,14 @@ NumericMatrix jaccard_coeff(NumericMatrix idx, bool weight) {
   return weights;
 }
 
+// [[Rcpp::export]]
+NumericMatrix jaccard_coeff(SEXP R_idx, SEXP R_weight) {
+  NumericMatrix idx(R_idx);
+  bool weight = as<bool>(R_weight);
+
+  return jaccard_coeff_cpp(idx, weight);
+}
+
 /***
  edges$C = jaccard_dist
  edges = subset(edges, C != 0)