Merge pull request #33 from COMBINE-lab/refactor

Refactor

Author: rob-p

CMakeLists.txt

@@ -4,15 +4,15 @@ enable_testing()
 
 project (RapMap)
 
-set(CPACK_PACKAGE_VERSION "0.3.0")
+set(CPACK_PACKAGE_VERSION "0.4.0")
 SET(CPACK_PACKAGE_VERSION_MAJOR "0")
-set(CPACK_PACKAGE_VERSION_MINOR "3")
+set(CPACK_PACKAGE_VERSION_MINOR "4")
 set(CPACK_PACKAGE_VERSION_PATCH "0")
 set(PROJECT_VERSION ${CPACK_PACKAGE_VERSION})
 set(CPACK_GENERATOR "TGZ")
 set(CPACK_SOURCE_GENERATOR "TGZ")
 set(CPACK_PACKAGE_VENDOR "Stony Brook University")
-set(CPACK_PACKAGE_DESCRIPTION_SUMMARY "RapMap - Wicked-fast quasi/pseudo/lightweight alignment")
+set(CPACK_PACKAGE_DESCRIPTION_SUMMARY "RapMap - Wicked-fast qasi-mapping")
 set(CPACK_PACKAGE_NAME
   "${CMAKE_PROJECT_NAME}-${CPACK_PACKAGE_VERSION_MAJOR}.${CPACK_PACKAGE_VERSION_MINOR}.${CPACK_PACKAGE_VERSION_PATCH}")
 set(CPACK_SOURCE_PACKAGE_FILE_NAME
@@ -26,8 +26,12 @@ set(CMAKE_MODULE_PATH ${CMAKE_MODULE_PATH} "${CMAKE_SOURCE_DIR}/cmake/Modules/")
 #    set(CMAKE_CXX_FLAGS "${CMAKE_CXX_FLAGS} -msse4.2 -DEMPHF_USE_POPCOUNT")
 #endif(SSE4_2_FOUND)
 
-set (WARNING_IGNORE_FLAGS "-Wno-deprecated-register -Wno-c++11-narrowing -Wno-unknown-pragmas")
-set (BOOST_CXX_FLAGS "-Wno-deprecated-register -std=c++11")
+if (APPLE)
+set (WARNING_IGNORE_FLAGS "-Wno-deprecated-register -Wno-unknon-pragmas -Wreturn-type -Werror=return-type")
+else()
+set (WARNING_IGNORE_FLAGS "-Wno-unknown-pragmas -Wreturn-type -Werror=return-type")
+endif()
+
 ## Prefer static to dynamic libraries
 SET(CMAKE_FIND_LIBRARY_SUFFIXES .a ${CMAKE_FIND_LIBRARY_SUFFIXES})
 
@@ -38,7 +42,14 @@ else()
   set (CMAKE_CXX_FLAGS "${CMAKE_CXX_FLAGS} -march=native")
 endif()
 
-set (CMAKE_CXX_FLAGS "${CMAKE_CXX_FLAGS} -pthread -funroll-loops -fPIC -fomit-frame-pointer -O4 -DHAVE_ANSI_TERM -Wall -std=c++11 -Wno-unknown-pragmas -Wreturn-type -Werror=return-type")
+
+if (QUIET_BUILD)
+  set(WALL "")
+else()
+  set(WALL "-Wall")
+endif()
+
+set (CMAKE_CXX_FLAGS "${CMAKE_CXX_FLAGS} -pthread -funroll-loops -fPIC -fomit-frame-pointer -O4 -DHAVE_ANSI_TERM ${WALL} -std=c++11")
 
 ##
 # OSX is strange (some might say, stupid in this regard).  Deal with it's quirkines here.
@@ -98,10 +109,6 @@ if ("${CMAKE_CXX_COMPILER_ID}" MATCHES "GNU")
     endif()
 
     set (WARNING_IGNORE_FLAGS "${WARNING_IGNORE_FLAGS} -Wno-unused-local-typedefs")
-    set (BOOST_TOOLSET "gcc")
-    set (BOOST_CONFIGURE_TOOLSET "--with-toolset=gcc")
-	set (BCXX_FLAGS "-std=c++11")
-    set (BOOST_EXTRA_FLAGS toolset=gcc cxxflags=${BCXX_FLAGS})
 # Tentatively, we support clang now
 elseif ("${CMAKE_CXX_COMPILER_ID}" MATCHES "Clang")
     set(CLANG TRUE)
@@ -111,10 +118,6 @@ elseif ("${CMAKE_CXX_COMPILER_ID}" MATCHES "Clang")
     if (HAVE_LIBCPP)
         message ("It appears that you're compiling with clang and that libc++ is available, so I'll use that")
         set (CMAKE_CXX_FLAGS "${CMAKE_CXX_FLAGS} -stdlib=libc++")
-	    set (BOOST_TOOLSET "clang")
-        set (BOOST_CONFIGURE_TOOLSET "--with-toolset=clang")
-	    set (BCXX_FLAGS "-stdlib=libc++ -DBOOST_HAS_INT128")
-	    set (BOOST_EXTRA_FLAGS toolset=clang cxxflags=${BCXX_FLAGS} linkflags="-stdlib=libc++")
         set (JELLYFISH_CXX_FLAGS "-stdlib=libc++")
     # Otherwise, use libstdc++ (and make it static)
     else()

README.md

@@ -89,3 +89,25 @@ RapMap is experimental, and the code, at this point, is subject to me testing ou
 # License 
 
 Since RapMap uses Jellyfish, it must be released under the GPL.  However, this is currently the only GPL dependency.  If it can be replaced, I'd like to re-license RapMap under the BSD license.  I'd be happy to accept pull-requests that replace the Jellyfish components with a library released under a more liberal license (BSD-compatible), but note that I will *not* accept such pull requests if they reduce the speed or increase the memory consumption over the Jellyfish-based version.
+
+# Citation
+
+If you use RapMap, or wish to cite the quasi-mapping concept or our algorithm for computing quasi-mappings, please 
+use this bibtex entry. 
+
+```
+@article{Srivastava15062016,
+author = {Srivastava, Avi and Sarkar, Hirak and Gupta, Nitish and Patro, Rob}, 
+title = {RapMap: a rapid, sensitive and accurate tool for mapping RNA-seq reads to transcriptomes},
+volume = {32}, 
+number = {12}, 
+pages = {i192-i200}, 
+year = {2016}, 
+doi = {10.1093/bioinformatics/btw277},
+URL = {http://bioinformatics.oxfordjournals.org/content/32/12/i192.abstract}, 
+eprint = {http://bioinformatics.oxfordjournals.org/content/32/12/i192.full.pdf+html}, 
+journal = {Bioinformatics} 
+}
+```
+
+Other citation formats for the RapMap paper are available [here](http://bioinformatics.oxfordjournals.org/citmgr?gca=bioinfo%3B32%2F12%2Fi192).

include/BooMap.hpp

@@ -52,6 +52,15 @@ class BooMap {
         data_.emplace_back(k, v);
     }
 
+    bool validate_hash(){
+        for( auto& e : data_ ) {
+            if (e.first != data_[boophf_->lookup(e.first)].first) {
+                std::cerr << "lookup of " << e.first << " failed!\n";
+            }
+        }
+        return true;
+    }
+
     bool build(int nthreads=1) {
         size_t numElem = data_.size();
         KeyIterator<decltype(data_.begin())> kb(data_.begin());
@@ -60,11 +69,8 @@ class BooMap {
         BooPHFT* ph = new BooPHFT(numElem, keyIt, nthreads);
         boophf_.reset(ph);
         std::cerr << "reordering keys and values to coincide with phf ... ";
-        std::vector<size_t> inds; inds.reserve(data_.size());
-        for (size_t i = 0; i < data_.size(); ++i) {
-            inds.push_back(ph->lookup(data_[i].first));
-        }
-        reorder_destructive_(inds.begin(), inds.end(), data_.begin());
+        reorder_fn_();
+        //validate_hash();
         std::cerr << "done\n";
         built_ = true;
         return built_;
@@ -164,6 +170,50 @@ class BooMap {
         return true;
     }
 
+  
+    void reorder_fn_()  {
+	/* Adapted from code at: http://blog.merovius.de/2014/08/12/applying-permutation-in-constant.html */
+        // Note, we can actually do this with out the bitvector by using the high-order bit 
+        // of the start of the suffix array intervals (since they are signed integers and negative
+        // positions are forbidden). 
+      std::vector<bool> bits(data_.size(), false);
+	for ( size_t i = 0; i < data_.size(); ++i ) {
+	  if (!bits[i]) {
+	    decltype(data_.front()) v = data_[i];
+	    auto j = boophf_->lookup(data_[i].first);
+	    while (i != j) {
+            auto pj = boophf_->lookup(data_[j].first);
+	      std::swap(data_[j], v);
+	      bits[j] = 1;
+	      j = pj; 
+	    }
+	    data_[i] = v;
+	  }
+	}
+
+	/* http://blog.merovius.de/2014/08/12/applying-permutation-in-constant.html
+	    for i := 0; i < len(vals); i++ {
+        if perm[i] < 0 {
+            // already correct - unmark and go on
+            // (note that ^a is the bitwise negation
+            perm[i] = ^perm[i]
+            continue
+        }
+
+        v, j := vals[i], perm[i]
+        for j != i {
+            vals[j], v = v, vals[j]
+            // When we find this element in the future, we must not swap it any
+            // further, so we mark it here
+            perm[j], j = ^perm[j], perm[j]
+        }
+        vals[i] = v
+    }
+}
+	*/
+    }
+
+
     // From : http://stackoverflow.com/questions/838384/reorder-vector-using-a-vector-of-indices
     template< typename order_iterator, typename value_iterator >
     void reorder_destructive_( order_iterator order_begin, order_iterator order_end, value_iterator v )  {

include/BooPHF.hpp

@@ -533,11 +533,11 @@ we need this 2-functors scheme because HashFunctors won't work with unordered_ma
 		//for debug purposes
 		void print() const
 		{
-			printf("bit array of size %lli: \n",_size);
+			printf("bit array of size %lu: \n",_size);
 			for(uint64_t ii = 0; ii< _size; ii++)
 			{
 				if(ii%10==0)
-					printf(" (%llu) ",ii);
+					printf(" (%lu) ",ii);
 				int val = (_bitArray[ii >> 6] >> (ii & 63 ) ) & 1;
 				printf("%i",val);
 			}
@@ -546,7 +546,7 @@ we need this 2-functors scheme because HashFunctors won't work with unordered_ma
 			printf("rank array : size %lu \n",_ranks.size());
 			for (uint64_t ii = 0; ii< _ranks.size(); ii++)
 			{
-				printf("%llu :  %lli,  ",ii,_ranks[ii]);
+				printf("%lu :  %lu,  ",ii,_ranks[ii]);
 			}
 			printf("\n");
 		}