Merge pull request #114 from EricSDavis/master

Add function for reading available normalizations from .hic file.

.gitignore

@@ -52,3 +52,4 @@ dist
 env
 hic_straw.egg-info
 
+.Rproj.user

R/DESCRIPTION

@@ -14,4 +14,4 @@ Encoding: UTF-8
 Imports: Rcpp
 LinkingTo: Rcpp
 NeedsCompilation: yes
-RoxygenNote: 7.1.1
+RoxygenNote: 7.2.0

R/NAMESPACE

@@ -2,6 +2,7 @@
 
 export(readHicBpResolutions)
 export(readHicChroms)
+export(readHicNormTypes)
 export(straw)
 import(Rcpp)
 useDynLib(strawr)

R/R/RcppExports.R

@@ -30,6 +30,17 @@ straw <- function(norm, fname, chr1loc, chr2loc, unit, binsize, matrix = "observ
     .Call('_strawr_straw', PACKAGE = 'strawr', norm, fname, chr1loc, chr2loc, unit, binsize, matrix)
 }
 
+#' Function for reading chromosomes from .hic file
+#'
+#' @param fname path to .hic file
+#' @return Data frame of chromosome names and lengths
+#' @examples
+#' readHicChroms(system.file("extdata", "test.hic", package = "strawr"))
+#' @export
+readHicChroms <- function(fname) {
+    .Call('_strawr_readHicChroms', PACKAGE = 'strawr', fname)
+}
+
 #' Function for reading basepair resolutions from .hic file
 #'
 #' @param fname path to .hic file
@@ -41,14 +52,14 @@ readHicBpResolutions <- function(fname) {
     .Call('_strawr_readHicBpResolutions', PACKAGE = 'strawr', fname)
 }
 
-#' Function for reading chromosomes from .hic file
-#'
+#' Function for reading available normalizations from .hic file
+#' 
 #' @param fname path to .hic file
-#' @return Data frame of chromosome names and lengths
+#' @return Vector of available normalizations
 #' @examples
-#' readHicChroms(system.file("extdata", "test.hic", package = "strawr"))
+#' readHicNormTypes(system.file("extdata", "test.hic", package = "strawr"))
 #' @export
-readHicChroms <- function(fname) {
-    .Call('_strawr_readHicChroms', PACKAGE = 'strawr', fname)
+readHicNormTypes <- function(fname) {
+    .Call('_strawr_readHicNormTypes', PACKAGE = 'strawr', fname)
 }
 

R/src/RcppExports.cpp

@@ -5,6 +5,11 @@
 
 using namespace Rcpp;
 
+#ifdef RCPP_USE_GLOBAL_ROSTREAM
+Rcpp::Rostream<true>&  Rcpp::Rcout = Rcpp::Rcpp_cout_get();
+Rcpp::Rostream<false>& Rcpp::Rcerr = Rcpp::Rcpp_cerr_get();
+#endif
+
 // straw
 Rcpp::DataFrame straw(std::string norm, std::string fname, std::string chr1loc, std::string chr2loc, const std::string& unit, int32_t binsize, std::string matrix);
 RcppExport SEXP _strawr_straw(SEXP normSEXP, SEXP fnameSEXP, SEXP chr1locSEXP, SEXP chr2locSEXP, SEXP unitSEXP, SEXP binsizeSEXP, SEXP matrixSEXP) {
@@ -22,6 +27,17 @@ BEGIN_RCPP
     return rcpp_result_gen;
 END_RCPP
 }
+// readHicChroms
+Rcpp::DataFrame readHicChroms(std::string fname);
+RcppExport SEXP _strawr_readHicChroms(SEXP fnameSEXP) {
+BEGIN_RCPP
+    Rcpp::RObject rcpp_result_gen;
+    Rcpp::RNGScope rcpp_rngScope_gen;
+    Rcpp::traits::input_parameter< std::string >::type fname(fnameSEXP);
+    rcpp_result_gen = Rcpp::wrap(readHicChroms(fname));
+    return rcpp_result_gen;
+END_RCPP
+}
 // readHicBpResolutions
 Rcpp::NumericVector readHicBpResolutions(std::string fname);
 RcppExport SEXP _strawr_readHicBpResolutions(SEXP fnameSEXP) {
@@ -33,22 +49,23 @@ BEGIN_RCPP
     return rcpp_result_gen;
 END_RCPP
 }
-// readHicChroms
-Rcpp::DataFrame readHicChroms(std::string fname);
-RcppExport SEXP _strawr_readHicChroms(SEXP fnameSEXP) {
+// readHicNormTypes
+Rcpp::CharacterVector readHicNormTypes(std::string fname);
+RcppExport SEXP _strawr_readHicNormTypes(SEXP fnameSEXP) {
 BEGIN_RCPP
     Rcpp::RObject rcpp_result_gen;
     Rcpp::RNGScope rcpp_rngScope_gen;
     Rcpp::traits::input_parameter< std::string >::type fname(fnameSEXP);
-    rcpp_result_gen = Rcpp::wrap(readHicChroms(fname));
+    rcpp_result_gen = Rcpp::wrap(readHicNormTypes(fname));
     return rcpp_result_gen;
 END_RCPP
 }
 
 static const R_CallMethodDef CallEntries[] = {
     {"_strawr_straw", (DL_FUNC) &_strawr_straw, 7},
-    {"_strawr_readHicBpResolutions", (DL_FUNC) &_strawr_readHicBpResolutions, 1},
     {"_strawr_readHicChroms", (DL_FUNC) &_strawr_readHicChroms, 1},
+    {"_strawr_readHicBpResolutions", (DL_FUNC) &_strawr_readHicBpResolutions, 1},
+    {"_strawr_readHicNormTypes", (DL_FUNC) &_strawr_readHicNormTypes, 1},
     {NULL, NULL, 0}
 };
 

R/src/straw.cpp

@@ -1291,21 +1291,23 @@ vector<chromosome> getChromosomes(string fname){
 //' Function for reading chromosomes from .hic file
 //'
 //' @param fname path to .hic file
-//' @return Data frame of chromosome names and lengths
+//' @return Data frame of chromosome indices, names and lengths
 //' @examples
 //' readHicChroms(system.file("extdata", "test.hic", package = "strawr"))
 //' @export
 // [[Rcpp::export]]
 Rcpp::DataFrame readHicChroms(std::string fname)
 {
   vector<chromosome> chroms = getChromosomes(std::move(fname));
+  Rcpp::NumericVector indices;
   Rcpp::StringVector names;
   Rcpp::NumericVector lengths;
   for (std::vector<chromosome>::iterator it = chroms.begin(); it != chroms.end(); ++it) {
+    indices.push_back(it->index);
     names.push_back(it->name);
     lengths.push_back(it->length);
   }
-  return Rcpp::DataFrame::create(Rcpp::Named("name") = names, Rcpp::Named("length") = lengths);
+  return Rcpp::DataFrame::create(Rcpp::Named("index") = indices, Rcpp::Named("name") = names, Rcpp::Named("length") = lengths);
 }
 
 //' Function for reading basepair resolutions from .hic file
@@ -1326,3 +1328,123 @@ Rcpp::NumericVector readHicBpResolutions(std::string fname)
   hiCFile->close();
   return bpResolutions;
 }
+
+// Reads all normalizations from the footer
+Rcpp::CharacterVector readNormsFromFooter(istream &fin, int64_t master, int32_t version) {
+
+    // Initialize variable to store norm types
+    Rcpp::CharacterVector normTypes;
+
+    // Read through the footer section
+    //nBytes
+    if (version > 8) {
+        readInt64FromFile(fin);
+    } else {
+        readInt32FromFile(fin);
+    }
+
+    // nEntries
+    int32_t nEntries = readInt32FromFile(fin);
+    for (int i = 0; i < nEntries; i++) {
+        string str;
+        getline(fin, str, '\0');
+        readInt64FromFile(fin); //fpos
+        readInt32FromFile(fin); //sizeInBytes
+    }
+
+    // nExpectedValues
+    int32_t nExpectedValues = readInt32FromFile(fin);
+    for (int i = 0; i < nExpectedValues; i++) {
+        string unit0;
+        getline(fin, unit0, '\0'); //unit
+        readInt32FromFile(fin);
+
+        int64_t nValues;
+        if (version > 8) {
+            nValues = readInt64FromFile(fin);
+            for (int j = 0; j < nValues; j++) {
+                readFloatFromFile(fin);
+            }
+        } else {
+            nValues = (int64_t) readInt32FromFile(fin);
+            for (int j = 0; j < nValues; j++) {
+                readDoubleFromFile(fin);
+            }
+        }
+
+        int32_t nNormalizationFactors = readInt32FromFile(fin);
+        for (int j = 0; j < nNormalizationFactors; j++) {
+            readInt32FromFile(fin); //chrIdx
+            if (version > 8) {
+                readFloatFromFile(fin); //v
+            } else {
+                readDoubleFromFile(fin);//v
+            }
+        }
+    }
+
+    // Needs to be read like this (readInt32FromFile doesn't work)
+    fin.read((char*)&nExpectedValues, sizeof(int32_t));
+    for (int i = 0; i < nExpectedValues; i++) {
+        //Record available norm types (handling empty strings as NONE)
+        string type;
+        getline(fin, type, '\0'); //typeString
+        if (type == "") {
+            type = "NONE";
+        }
+        normTypes.push_back(type);
+
+        string unit0;
+        getline(fin, unit0, '\0'); //unit
+        readInt32FromFile(fin);
+
+        int64_t nValues;
+        if (version > 8) {
+            nValues = readInt64FromFile(fin);
+            for (int j = 0; j < nValues; j++) {
+                readFloatFromFile(fin); //v
+            }
+        } else {
+            nValues = (int64_t) readInt32FromFile(fin);
+            for (int j = 0; j < nValues; j++) {
+                readDoubleFromFile(fin); //v
+            }
+        }
+
+        int32_t nNormalizationFactors = readInt32FromFile(fin);
+        for (int j = 0; j < nNormalizationFactors; j++) {
+            readInt32FromFile(fin); //chrIdx
+            if (version > 8) {
+                readFloatFromFile(fin); //v
+            } else {
+                readDoubleFromFile(fin); //v
+            }
+        }
+    }
+
+    // Include "NONE"
+    normTypes.push_back("NONE");
+
+    // Return unique norms
+    return unique(normTypes);
+}
+
+//' Function for reading available normalizations from .hic file
+//' 
+//' @param fname path to .hic file
+//' @return Vector of available normalizations
+//' @examples
+//' readHicNormTypes(system.file("extdata", "test.hic", package = "strawr"))
+//' @export
+// [[Rcpp::export]]
+Rcpp::CharacterVector readHicNormTypes(std::string fname)
+{
+    HiCFile *hiCFile = new HiCFile(std::move(fname));
+    Rcpp::CharacterVector normTypes;
+    hiCFile->fin.seekg(hiCFile->master, ios::beg);
+    normTypes = readNormsFromFooter(hiCFile->fin,
+                                    hiCFile->master,
+                                    hiCFile->version);
+    hiCFile->close();
+    return normTypes;
+}