biology/haplohseq: Update to 0.1.2.3

Unbreak build for boost 1.87+ (hayzam@alchemilla.io)
Switch master site to fork on Github and upstream patches
MD Anderson has ceased development and is planning a replacement

PR:             286955
Reported by:    hayzam@alchemilla.io

8 files changed, 7 insertions, 246 deletions

diff --git a/biology/haplohseq/Makefile b/biology/haplohseq/Makefile
index 8da67ec49753..128b5317242d 100644
--- a/biology/haplohseq/Makefile
+++ b/biology/haplohseq/Makefile
@@ -1,16 +1,15 @@
 PORTNAME=	haplohseq
-DISTVERSION=	0.1.2
-PORTREVISION=	13
+DISTVERSION=	0.1.2-3
+DISTVERSIONSUFFIX=	-gfeb9267
 CATEGORIES=	biology
 
 MAINTAINER=	jwb@FreeBSD.org
 COMMENT=	Identify regions of allelic imbalance
-WWW=		https://sites.google.com/site/integrativecancergenomics/software/haplohseq
+WWW=		https://github.com/outpaddling/haplohseq \
+		https://sites.google.com/site/integrativecancergenomics/software/haplohseq
 
 LICENSE=	MIT
 
-BROKEN=		fails to build with Boost>=1.87
-
 LIB_DEPENDS=	libboost_system.so:devel/boost-libs
 RUN_DEPENDS=	R-cran-optparse>0:devel/R-cran-optparse
 
diff --git a/biology/haplohseq/distinfo b/biology/haplohseq/distinfo
index 7af7556d5596..ab9872a8d79f 100644
--- a/biology/haplohseq/distinfo
+++ b/biology/haplohseq/distinfo
@@ -1,3 +1,3 @@
-TIMESTAMP = 1611421648
-SHA256 (outpaddling-haplohseq-0.1.2_GH0.tar.gz) = 64f61023a6795decb680c23d142b1a349988d4c0e3bef8c7d52ca33bf42f29f0
-SIZE (outpaddling-haplohseq-0.1.2_GH0.tar.gz) = 12084603
+TIMESTAMP = 1747762325
+SHA256 (outpaddling-haplohseq-0.1.2-3-gfeb9267_GH0.tar.gz) = a23691f16f1c03e0dfd833896068e05eaa139d488a7094a49033aeacba948cb1
+SIZE (outpaddling-haplohseq-0.1.2-3-gfeb9267_GH0.tar.gz) = 12085111
diff --git a/biology/haplohseq/files/patch-makefile b/biology/haplohseq/files/patch-makefile
deleted file mode 100644
index 5387716aa503..000000000000
--- a/biology/haplohseq/files/patch-makefile
+++ /dev/null
@@ -1,105 +0,0 @@
---- makefile.orig	2019-08-29 18:03:23 UTC
-+++ makefile
-@@ -7,23 +7,32 @@
- #
- 
- # Constants
--CXX=g++
-+# Default to g++ if not set by make args or environment
-+CXX?=g++
- #-O0 -g will turn on debugging
- #The rule of thumb:
- #When you need to debug, use -O0 (and -g to generate debugging symbols.)
- #When you are preparing to ship it, use -O2.
- #When you use gentoo, use -O3...!
- #When you need to put it on an embedded system, use -Os (optimize for size, not for efficiency.)
--CXX_FLAGS=-Wall -g -stdlib=libstdc++
-+# Use canonincal compiler variables, which may be provided by build env
-+CXXFLAGS?=-Wall -g -stdlib=libstdc++
- SRC=src
- CONF=conf
- HMM_SRC=$(SRC)/hmm
- UTIL_SRC=$(SRC)/util
--BOOST=/usr/local/boost_1_52_0
-+BOOST?=/usr/local/boost_1_52_0
- INCLUDES=-I./$(SRC) -I./$(HMM_SRC) -I./$(UTIL_SRC) -isystem$(BOOST)
--LIBRARY_PATHS=-Llib/macosx 
-+LIBRARY_PATHS?=-Llib/macosx
- LIBRARIES=-lm -lboost_program_options -lboost_system -lboost_filesystem -lboost_thread
- 
-+# Installation target with destdir support
-+DESTDIR?=.
-+PREFIX?=/usr/local
-+MKDIR?=mkdir
-+INSTALL?=install
-+STRIP?= # empty, set to -s to install stripped binary
-+
- # Generated directories which are generated in this script and cleaned up with 'make clean'
- BUILD=build
- OBJ=$(BUILD)/obj
-@@ -35,18 +44,21 @@ directories:
- 
- # Create object files into the OBJ directory from cpp files in the SRC directory.
- $(OBJ)/%.o:	$(SRC)/%.cpp directories
--	$(CXX) $(CXX_FLAGS) $(INCLUDES) -c -o $@ $<
-+	$(CXX) $(CXXFLAGS) $(INCLUDES) -c -o $@ $<
- $(OBJ)/%.o:	$(HMM_SRC)/%.cpp directories
--	$(CXX) $(CXX_FLAGS) $(INCLUDES) -c -o $@ $<
-+	$(CXX) $(CXXFLAGS) $(INCLUDES) -c -o $@ $<
- $(OBJ)/%.o:	$(UTIL_SRC)/%.cpp directories
--	$(CXX) $(CXX_FLAGS) $(INCLUDES) -c -o $@ $<
-+	$(CXX) $(CXXFLAGS) $(INCLUDES) -c -o $@ $<
- 
--all: clean haplohseq
-+all: haplohseq test
- 
- haplohseq: $(OBJ)/Hmm.o $(OBJ)/DataStructures.o $(OBJ)/MathUtil.o $(OBJ)/StringUtil.o $(OBJ)/InputProcessor.o $(OBJ)/Reporter.o $(OBJ)/FreqPhase.o $(OBJ)/VcfUtil.o $(OBJ)/HaplohSeq.o
--	$(CXX) -o $(BIN)/$@ $(CXX_FLAGS) $(INCLUDES) $(OBJ)/Hmm.o $(OBJ)/DataStructures.o $(OBJ)/MathUtil.o $(OBJ)/StringUtil.o $(OBJ)/InputProcessor.o $(OBJ)/Reporter.o $(OBJ)/FreqPhase.o $(OBJ)/VcfUtil.o $(OBJ)/HaplohSeq.o $(LIBRARY_PATHS) $(LIBRARIES) 
--	cp -r $(CONF) $(BIN)/.
--	
-+	$(CXX) -o $(BIN)/$@ $(CXXFLAGS) $(INCLUDES) $(OBJ)/Hmm.o $(OBJ)/DataStructures.o $(OBJ)/MathUtil.o $(OBJ)/StringUtil.o $(OBJ)/InputProcessor.o $(OBJ)/Reporter.o $(OBJ)/FreqPhase.o $(OBJ)/VcfUtil.o $(OBJ)/HaplohSeq.o $(LIBRARY_PATHS) $(LIBRARIES) $(LDFLAGS)
-+
-+install:
-+	$(MKDIR) -p $(DESTDIR)$(PREFIX)/bin
-+	$(INSTALL) -c $(STRIP) $(BIN)/* $(DESTDIR)$(PREFIX)/bin
-+
- clean:
- 	rm -rf $(BUILD)
- 
-@@ -74,28 +86,28 @@ test: FreqPhaseTest HaplohSeqTest HmmTest InputProcess
- 
- # Create test object files into the OBJ directory from cpp files in the SRC directory.
- $(TEST_BIN)/%.o:	$(TEST_SRC)/%.cpp directories test_directory
--	$(CXX) $(CXX_FLAGS) $(INCLUDES) $(TEST_INCLUDES) -c -o $@ $<
-+	$(CXX) $(CXXFLAGS) $(INCLUDES) $(TEST_INCLUDES) -c -o $@ $<
- 
- HaplohSeqTest:	$(OBJ)/Hmm.o $(OBJ)/DataStructures.o $(OBJ)/MathUtil.o $(OBJ)/StringUtil.o $(OBJ)/InputProcessor.o $(TEST_BIN)/HaplohSeqTest.o
--	$(CXX) -o $(TEST_BIN)/$@ $(CXX_FLAGS) $(OBJ)/Hmm.o $(OBJ)/DataStructures.o $(OBJ)/MathUtil.o $(OBJ)/StringUtil.o $(OBJ)/InputProcessor.o $(TEST_BIN)/HaplohSeqTest.o $(LIBRARY_PATHS) $(LIBRARIES) 
-+	$(CXX) -o $(TEST_BIN)/$@ $(CXXFLAGS) $(OBJ)/Hmm.o $(OBJ)/DataStructures.o $(OBJ)/MathUtil.o $(OBJ)/StringUtil.o $(OBJ)/InputProcessor.o $(TEST_BIN)/HaplohSeqTest.o $(LIBRARY_PATHS) $(LIBRARIES) $(LDFLAGS)
- 
- FreqPhaseTest:	$(OBJ)/StringUtil.o $(OBJ)/InputProcessor.o $(OBJ)/FreqPhase.o $(TEST_BIN)/FreqPhaseTest.o
--	$(CXX) -o $(TEST_BIN)/$@ $(CXX_FLAGS) $(OBJ)/StringUtil.o $(OBJ)/InputProcessor.o $(OBJ)/FreqPhase.o $(TEST_BIN)/FreqPhaseTest.o $(LIBRARY_PATHS) $(LIBRARIES) 
-+	$(CXX) -o $(TEST_BIN)/$@ $(CXXFLAGS) $(OBJ)/StringUtil.o $(OBJ)/InputProcessor.o $(OBJ)/FreqPhase.o $(TEST_BIN)/FreqPhaseTest.o $(LIBRARY_PATHS) $(LIBRARIES) $(LDFLAGS)
- 
- HmmTest:	$(OBJ)/Hmm.o $(OBJ)/DataStructures.o $(OBJ)/MathUtil.o $(OBJ)/StringUtil.o $(TEST_BIN)/HmmTest.o
--	$(CXX) -o $(TEST_BIN)/$@ $(CXX_FLAGS) $(OBJ)/Hmm.o $(OBJ)/DataStructures.o $(OBJ)/MathUtil.o $(OBJ)/StringUtil.o $(TEST_BIN)/HmmTest.o $(LIBRARY_PATHS) $(LIBRARIES) 
-+	$(CXX) -o $(TEST_BIN)/$@ $(CXXFLAGS) $(OBJ)/Hmm.o $(OBJ)/DataStructures.o $(OBJ)/MathUtil.o $(OBJ)/StringUtil.o $(TEST_BIN)/HmmTest.o $(LIBRARY_PATHS) $(LIBRARIES) $(LDFLAGS)
- 
- InputProcessorTest:	$(OBJ)/Hmm.o $(OBJ)/DataStructures.o $(OBJ)/MathUtil.o $(OBJ)/StringUtil.o $(OBJ)/InputProcessor.o $(TEST_BIN)/InputProcessorTest.o
--	$(CXX) -o $(TEST_BIN)/$@ $(CXX_FLAGS) $(OBJ)/Hmm.o $(OBJ)/DataStructures.o $(OBJ)/MathUtil.o $(OBJ)/StringUtil.o $(OBJ)/InputProcessor.o $(TEST_BIN)/InputProcessorTest.o $(LIBRARY_PATHS) $(LIBRARIES) 
-+	$(CXX) -o $(TEST_BIN)/$@ $(CXXFLAGS) $(OBJ)/Hmm.o $(OBJ)/DataStructures.o $(OBJ)/MathUtil.o $(OBJ)/StringUtil.o $(OBJ)/InputProcessor.o $(TEST_BIN)/InputProcessorTest.o $(LIBRARY_PATHS) $(LIBRARIES) $(LDFLAGS)
- 	
- MathUtilTest:	$(OBJ)/Hmm.o $(OBJ)/DataStructures.o $(OBJ)/MathUtil.o $(OBJ)/StringUtil.o $(TEST_BIN)/MathUtilTest.o
--	$(CXX) -o $(TEST_BIN)/$@ $(CXX_FLAGS) $(OBJ)/Hmm.o $(OBJ)/DataStructures.o $(OBJ)/MathUtil.o $(OBJ)/StringUtil.o $(TEST_BIN)/MathUtilTest.o $(LIBRARY_PATHS) $(LIBRARIES) 
-+	$(CXX) -o $(TEST_BIN)/$@ $(CXXFLAGS) $(OBJ)/Hmm.o $(OBJ)/DataStructures.o $(OBJ)/MathUtil.o $(OBJ)/StringUtil.o $(TEST_BIN)/MathUtilTest.o $(LIBRARY_PATHS) $(LIBRARIES) $(LDFLAGS)
- 	
- ReporterTest:	$(OBJ)/Hmm.o $(OBJ)/DataStructures.o $(OBJ)/MathUtil.o $(OBJ)/StringUtil.o $(TEST_BIN)/ReporterTest.o
--	$(CXX) -o $(TEST_BIN)/$@ $(CXX_FLAGS) $(OBJ)/Hmm.o $(OBJ)/DataStructures.o $(OBJ)/MathUtil.o $(OBJ)/StringUtil.o $(TEST_BIN)/ReporterTest.o $(LIBRARY_PATHS) $(LIBRARIES) 
-+	$(CXX) -o $(TEST_BIN)/$@ $(CXXFLAGS) $(OBJ)/Hmm.o $(OBJ)/DataStructures.o $(OBJ)/MathUtil.o $(OBJ)/StringUtil.o $(TEST_BIN)/ReporterTest.o $(LIBRARY_PATHS) $(LIBRARIES) $(LDFLAGS)
- 
- ThreadPoolTest:	$(OBJ)/Hmm.o $(OBJ)/DataStructures.o $(OBJ)/MathUtil.o $(OBJ)/StringUtil.o $(OBJ)/ThreadPool.o $(TEST_BIN)/ThreadPoolTest.o
--	$(CXX) -o $(TEST_BIN)/$@ $(CXX_FLAGS) $(OBJ)/Hmm.o $(OBJ)/DataStructures.o $(OBJ)/MathUtil.o $(OBJ)/StringUtil.o $(OBJ)/ThreadPool.o $(TEST_BIN)/ThreadPoolTest.o $(LIBRARY_PATHS) $(LIBRARIES)
-+	$(CXX) -o $(TEST_BIN)/$@ $(CXXFLAGS) $(OBJ)/Hmm.o $(OBJ)/DataStructures.o $(OBJ)/MathUtil.o $(OBJ)/StringUtil.o $(OBJ)/ThreadPool.o $(TEST_BIN)/ThreadPoolTest.o $(LIBRARY_PATHS) $(LIBRARIES) $(LDFLAGS)
- ############################## END TEST LOGIC ##############################
- 
- 
diff --git a/biology/haplohseq/files/patch-scripts_ldmap.py b/biology/haplohseq/files/patch-scripts_ldmap.py
deleted file mode 100644
index f9fb588048b1..000000000000
--- a/biology/haplohseq/files/patch-scripts_ldmap.py
+++ /dev/null
@@ -1,42 +0,0 @@
---- scripts/ldmap.py.orig	2021-01-23 17:32:15 UTC
-+++ scripts/ldmap.py
-@@ -29,10 +29,10 @@ class MapMaker:
-     coordinates of the haplotype marker genomic positions.
-     '''
-     def __init__(self, markerFilename, haplotypesFilename):
--        print "reading haplotype base coordinates ...."
-+        print("reading haplotype base coordinates ....")
-         self.coords, self.coordIndexes, self.refs, self.alts = self.readMarkers(markerFilename)
- #         self.haplotypes = self.readHaplotypes(haplotypesFilename)
--        print "reading haplotype panel ...."
-+        print("reading haplotype panel ....")
-         self.markerCalls = self.readHaplotypes(haplotypesFilename, transpose = True)
-     
-     # reads from a vcf file of defined positions representing the haplotype data    
-@@ -118,7 +118,7 @@ class MapMaker:
-         # 1) COLLECT MARGINALS AND IDENTIFY SITES THAT ARE POLYMORPHIC
-         # If a site has only one allele in the haplotype panel it is not informative, so it won't be used
-         # in LD calculations.  We may want to consider setting some minimum threshold for a minor allele frequency.
--        print "collecting marginal allele counts and identifying polymorphic sites from haplotype panel ...."
-+        print("collecting marginal allele counts and identifying polymorphic sites from haplotype panel ....")
-         for index in range(0,len(self.markerCalls)):
-             ref = self.refs[index]
-             alt = self.alts[index]
-@@ -152,7 +152,7 @@ class MapMaker:
-         log.write("Num non-polymorphic sites: " + str(numNonpolymorphicSites) + "\n")
-                
-         # 2) CALCULATE D FOR POLYMORPHIC SITES COMPARED WITH N SITES TO THE RIGHT THAT ARE POLYMORPHIC
--        print "calculating D between polymorphic sites and their neighbors ...."
-+        print("calculating D between polymorphic sites and their neighbors ....")
-         dVals = []
-         for i in range(0, len(polymorphicIndexes)):   # the last coordinate won't have any partners
-             index_i = polymorphicIndexes[i]
-@@ -244,7 +244,7 @@ class LDMap:
-         return ldMap
-         
-     def save(self, filename):
--        print "saving ldmap to " + filename
-+        print("saving ldmap to " + filename)
-         ldMapFile = open(filename, 'w')
-         ldMapFile.write("\t".join(["COORD","REF","ALT","PAIRED_ALLELES"]) + "\n")
-         for i in range(0,len(self.coords)):
diff --git a/biology/haplohseq/files/patch-scripts_simple__phaser.py b/biology/haplohseq/files/patch-scripts_simple__phaser.py
deleted file mode 100644
index c6c5941ee00d..000000000000
--- a/biology/haplohseq/files/patch-scripts_simple__phaser.py
+++ /dev/null
@@ -1,31 +0,0 @@
---- scripts/simple_phaser.py.orig	2021-01-23 17:25:56 UTC
-+++ scripts/simple_phaser.py
-@@ -108,7 +108,7 @@ class SimplePhaser:
-         return informativeHets, informativeCoords, informativeIndexes
-     
-     def getRefPairedAllele(self, anchorIndex, pairedIndex):
--#         print "\t".join([str(anchorIndex), str(pairedIndex - anchorIndex  - 1), str(len(self.ldMap.dVals)),str(len(self.ldMap.dVals[anchorIndex]))]) # prints intermarker counts
-+#         print("\t".join([str(anchorIndex), str(pairedIndex - anchorIndex  - 1), str(len(self.ldMap.dVals)),str(len(self.ldMap.dVals[anchorIndex]))]) # prints intermarker counts)
-         if (pairedIndex - anchorIndex - 1) >= len(self.ldMap.dVals[anchorIndex]):
-             pRef = random.uniform(0,1)
-             if pRef > 0.5:
-@@ -166,8 +166,8 @@ class SimplePhaser:
-                 altHap = temp 
-             
-             if pairedAllele not in [ref,alt]:
--                print "pairedAllele not ref or alt " + ref + " " + alt
--                print pairedAllele               
-+                print("pairedAllele not ref or alt " + ref + " " + alt)
-+                print(pairedAllele)
-          
-         return phasedAlleles, informativeCoords
-             
-@@ -182,7 +182,7 @@ class SimplePhaser:
-             elif candidateHap[i].upper() == phasedHaps[1][i].upper():
-                 currentMatch = 1
-             else:
--                print "ERROR: " + candidateHap[i] + " does not match " + phasedHaps[0][i] + " or " + phasedHaps[1][i] + "\t" + self.ldMap.coords[i] + "\t" + self.ldMap.refs[i] + "\t" + self.ldMap.alts[i]
-+                print("ERROR: " + candidateHap[i] + " does not match " + phasedHaps[0][i] + " or " + phasedHaps[1][i] + "\t" + self.ldMap.coords[i] + "\t" + self.ldMap.refs[i] + "\t" + self.ldMap.alts[i])
- #                 # randomly assign match
- #                 p0 = random.uniform(0,1)
- #                 if p0 > 0.5:
diff --git a/biology/haplohseq/files/patch-src_FreqPhase.cpp b/biology/haplohseq/files/patch-src_FreqPhase.cpp
deleted file mode 100644
index 9bab76aee2f5..000000000000
--- a/biology/haplohseq/files/patch-src_FreqPhase.cpp
+++ /dev/null
@@ -1,29 +0,0 @@
---- src/FreqPhase.cpp.orig	2019-11-13 14:19:06 UTC
-+++ src/FreqPhase.cpp
-@@ -5,6 +5,7 @@
-  * Email:  sanlucas@gmail.com
-  */
- 
-+#include <sysexits.h>
- #include "FreqPhase.h"
- 
- namespace haplohseq {
-@@ -180,10 +181,17 @@ double FreqPhase::meanValue(const std::v
- double FreqPhase::medianValue(const std::vector<double>& values) {
- 	double median;
- 	size_t size = values.size();
-+
-+	// We can probably detect this condition earlier while loading the VCF
-+	if ( size == 0 ) {
-+		std::cerr << "FreqPhase::medianValue(): values vector is empty." << std::endl;
-+		std::cerr << "Make sure your VCF has all of GT:AD:DP in the FORMAT column." << std::endl;
-+		exit(EX_DATAERR);
-+	}
- 	std::vector<double> tempFreqs(values);
- 	sort(tempFreqs.begin(), tempFreqs.end());
- 
--	if (size  % 2 == 0) {
-+	if (size % 2 == 0) {
- 		median = (tempFreqs[size / 2 - 1] + tempFreqs[size / 2]) / 2;
- 	}
- 	else {
diff --git a/biology/haplohseq/files/patch-src_HaplohSeq.cpp b/biology/haplohseq/files/patch-src_HaplohSeq.cpp
deleted file mode 100644
index a80ce577f747..000000000000
--- a/biology/haplohseq/files/patch-src_HaplohSeq.cpp
+++ /dev/null
@@ -1,20 +0,0 @@
---- src/HaplohSeq.cpp.orig	2019-09-01 14:58:30 UTC
-+++ src/HaplohSeq.cpp
-@@ -202,7 +202,7 @@ void HaplohSeq::runBafHaplohseq(	std::string& obsType,
- 
- 		// Spawn worker threads
- 		for (std::size_t t = 0; t < numThreads; t++) {
--			threadPool.create_thread(boost::bind((unsigned long int (boost::asio::io_service::*)())&boost::asio::io_service::run, &ioService));
-+			threadPool.create_thread(boost::bind((boost::asio::io_context::count_type (boost::asio::io_service::*)())&boost::asio::io_service::run, &ioService));
- 		}
- 
- 		std::map<std::string, boost::shared_ptr<Hmm> > chrHmms;
-@@ -420,7 +420,7 @@ void HaplohSeq::runVcfHaplohseq(	std::string& obsType,
- 
- 		// Spawn worker threads
- 		for (std::size_t t = 0; t < numThreads; t++) {
--			threadPool.create_thread(boost::bind((unsigned long int (boost::asio::io_service::*)())&boost::asio::io_service::run, &ioService));
-+			threadPool.create_thread(boost::bind((boost::asio::io_context::count_type (boost::asio::io_service::*)())&boost::asio::io_service::run, &ioService));
- 		}
- 
- 		std::map<std::string, boost::shared_ptr<Hmm> > chrHmms;
diff --git a/biology/haplohseq/files/patch-src_test_InputProcessorTest.cpp b/biology/haplohseq/files/patch-src_test_InputProcessorTest.cpp
deleted file mode 100644
index 21ea56a8f600..000000000000
--- a/biology/haplohseq/files/patch-src_test_InputProcessorTest.cpp
+++ /dev/null
@@ -1,11 +0,0 @@
---- src/test/InputProcessorTest.cpp.orig	2019-09-01 12:33:22 UTC
-+++ src/test/InputProcessorTest.cpp
-@@ -82,7 +82,7 @@ BOOST_AUTO_TEST_CASE(isHetTest) {
- 	BOOST_REQUIRE_EQUAL(inputProc.isHet('A','?'), false);
- 	BOOST_REQUIRE_EQUAL(inputProc.isHet('A','.'), false);
- 	BOOST_REQUIRE_EQUAL(inputProc.isHet('A',' '), false);
--	BOOST_REQUIRE_EQUAL(inputProc.isHet('C', NULL), true);
-+	BOOST_REQUIRE_EQUAL(inputProc.isHet('C','\0'), true);
- }
- 
- BOOST_AUTO_TEST_CASE(readAllelesTest) {