gencodegenes 1.2.0__tar.gz → 1.2.1__tar.gz

{gencodegenes-1.2.0 → gencodegenes-1.2.1}/MANIFEST.in

@@ -10,8 +10,10 @@ include src/gencodegenes/py.typed
 include src/*.h
 include src/*.cpp
 
-include src/zlib/*.c
-include src/zlib/*.h
+graft src/zlib-ng
+prune src/zlib-ng/test
+prune src/zlib-ng/.github
+exclude src/zlib-ng/.git
 
 # gencodegenes tests
 include tests/*.py

{gencodegenes-1.2.0/src/gencodegenes.egg-info → gencodegenes-1.2.1}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: gencodegenes
-Version: 1.2.0
+Version: 1.2.1
 Summary: Package to load genes from GENCODE GTF files
 Author-email: Jeremy McRae <jeremy.mcrae@gmail.com>
 License-Expression: MIT

{gencodegenes-1.2.0 → gencodegenes-1.2.1}/pyproject.toml

@@ -1,10 +1,10 @@
 [build-system]
-requires = ["cython", "setuptools >= 77.0.0"]
+requires = ["cython", "setuptools >= 77.0.0", "cmake >= 3.14"]
 build-backend = "setuptools.build_meta"
 
 [project]
 name = 'gencodegenes'
-version = '1.2.0'
+version = '1.2.1'
 description = 'Package to load genes from GENCODE GTF files'
 readme = 'README.md'
 license = "MIT"

gencodegenes-1.2.1/setup.py

@@ -0,0 +1,99 @@
+
+import os
+import subprocess
+import sys
+import shutil
+
+from setuptools import setup, Extension
+from setuptools.command.build_ext import build_ext
+from Cython.Build import cythonize
+
+EXTRA_COMPILE_ARGS = []
+EXTRA_LINK_ARGS = []
+if sys.platform == "win32":
+    EXTRA_COMPILE_ARGS += ['/std:c++14']
+else:
+    EXTRA_COMPILE_ARGS += ['-std=c++11']
+    if sys.platform == "darwin":
+        EXTRA_COMPILE_ARGS += ["-stdlib=libc++"]
+        EXTRA_LINK_ARGS += ["-stdlib=libc++"]
+        sdk_base = "/Library/Developer/CommandLineTools/SDKs/MacOSX.sdk"
+        if os.path.exists(sdk_base):
+            EXTRA_COMPILE_ARGS += [
+                f"-I{sdk_base}/usr/include/c++/v1",
+                f"-I{sdk_base}/usr/include",
+            ]
+            EXTRA_LINK_ARGS += [
+                f"-L{sdk_base}/usr/lib",
+            ]
+
+ZLIB_NG_DIR = os.path.abspath('src/zlib-ng')
+
+class build_ext_zlib_ng(build_ext):
+    ''' build zlib-ng as a static library (via cmake), before building extensions
+    
+    zlib-ng is built with its native API (zng_ prefixed functions), so it can't
+    clash with any system zlib loaded into the python process.
+    '''
+    def run(self):
+        build_dir = os.path.abspath(os.path.join(self.build_temp, 'zlib-ng'))
+        config = [
+            '-DCMAKE_BUILD_TYPE=Release',
+            '-DCMAKE_POSITION_INDEPENDENT_CODE=ON',
+            '-DBUILD_SHARED_LIBS=OFF',
+            '-DZLIB_COMPAT=OFF',
+            '-DWITH_GZFILEOP=ON',
+            '-DBUILD_TESTING=OFF',
+        ]
+        archflags = os.environ.get('ARCHFLAGS', '')
+        if sys.platform == 'darwin' and archflags:
+            # match the architectures requested for the extension (e.g. by cibuildwheel)
+            archs = ';'.join(archflags.split()[1::2])
+            config.append(f'-DCMAKE_OSX_ARCHITECTURES={archs}')
+        subprocess.check_call(['cmake', '-S', ZLIB_NG_DIR, '-B', build_dir] + config)
+        subprocess.check_call(['cmake', '--build', build_dir, '--config', 'Release',
+            '--parallel'])
+        
+        lib_dir = build_dir
+        lib = 'z-ng'
+        if sys.platform == 'win32':
+            lib_dir = os.path.join(build_dir, 'Release')
+            lib = 'zlibstatic-ng'
+        for ext in self.extensions:
+            if ext.name == 'gencodegenes.gencode':
+                ext.include_dirs.append(build_dir)
+                ext.library_dirs.append(lib_dir)
+                ext.libraries.append(lib)
+        super().run()
+
+extensions = [
+    Extension("gencodegenes.transcript",
+        extra_compile_args=EXTRA_COMPILE_ARGS,
+        extra_link_args=EXTRA_LINK_ARGS,
+        sources=[
+            "src/gencodegenes/transcript.pyx",
+            "src/tx.cpp"],
+        include_dirs=["src/"],
+        language="c++"),
+    Extension("gencodegenes.gencode",
+        extra_compile_args=EXTRA_COMPILE_ARGS,
+        extra_link_args=EXTRA_LINK_ARGS,
+        sources=[
+            "src/gencodegenes/gencode.pyx",
+            "src/gencode.cpp",
+            "src/gtf.cpp",
+            "src/tx.cpp"],
+        include_dirs=["src/"],
+        define_macros=[('WITH_GZFILEOP', None)],
+        language="c++"),
+    ]
+
+# include tx.h in the package, for downstream usage
+shutil.copy("src/tx.h", "src/gencodegenes/tx.h")
+shutil.copy("src/tx.cpp", "src/gencodegenes/tx.cpp")
+
+setup(
+    package_dir={'': 'src'},
+    ext_modules=cythonize(extensions),
+    cmdclass={'build_ext': build_ext_zlib_ng},
+    )

{gencodegenes-1.2.0 → gencodegenes-1.2.1}/src/gencode.cpp

@@ -7,6 +7,7 @@
 #include <set>
 #include <stdexcept>
 #include <unordered_map>
+#include <utility>
 #include <vector>
 
 #include <iostream>
@@ -18,18 +19,18 @@
 namespace gencode {
 
 // check which exon is first, by start position
-static bool compareExons(std::vector<int> e1, std::vector<int> e2) {
-    return (e1[0] < e2[0]);
+static bool compareExons(const Region & e1, const Region & e2) {
+    return (e1.start < e2.start);
 }
 
-static void sort_exons(std::vector<std::vector<int> > & exons) {
+static void sort_exons(std::vector<Region> & exons) {
     std::sort(exons.begin(), exons.end(), compareExons);
 }
 
 // find the index of the exon containing a given chromosome position
-static std::uint32_t get_exon_num(std::vector<std::vector<int> > exons, int pos) {
+static std::uint32_t get_exon_num(const std::vector<Region> & exons, int pos) {
     for (std::uint32_t i=0; i<exons.size(); i++) {
-        if ((pos >= exons[i][0]) && (pos <= exons[i][1])) {
+        if ((pos >= exons[i].start) && (pos <= exons[i].end)) {
             return i;
         }
     }
@@ -43,34 +44,33 @@ static std::uint32_t get_exon_num(std::vector<std::vector<int> > exons, int pos)
 // just set the first CDS coord and last CDS coord to their values though,
 // as at least one stop codon spans an intron boundary, which messes up the
 // CDS if included as is.
-static void include_end_codons(std::map<std::string, int> cds_range, TxInfo & info) {
+static void include_end_codons(const std::map<std::string, int> & cds_range, TxInfo & info) {
     if (info.cds.size() == 0) {
         return;
     }
     sort_exons(info.cds);
     sort_exons(info.exons);
+    int cds_min = cds_range.at("min");
+    int cds_max = cds_range.at("max");
 
     // handle left (5') boundary
-    std::uint32_t first_idx = get_exon_num(info.exons, info.cds[0][0]);
-    std::uint32_t min_idx = get_exon_num(info.exons, cds_range["min"]);
+    std::uint32_t first_idx = get_exon_num(info.exons, info.cds[0].start);
+    std::uint32_t min_idx = get_exon_num(info.exons, cds_min);
     if (min_idx == first_idx) {
-        info.cds[0][0] = cds_range["min"];
+        info.cds[0].start = cds_min;
     } else {
-        info.cds[0][0] = info.exons[first_idx][0];  // extend existing CDS
-        std::vector<int> extra_cds = {cds_range["min"], info.exons[min_idx][1]};
-        auto it = info.cds.begin();
-        info.cds.insert(it, extra_cds);
+        info.cds[0].start = info.exons[first_idx].start;  // extend existing CDS
+        info.cds.insert(info.cds.begin(), Region {cds_min, info.exons[min_idx].end});
     }
 
     // handle right (3') boundary
-    std::uint32_t last_idx = get_exon_num(info.exons, info.cds.back()[1]);
-    std::uint32_t max_idx = get_exon_num(info.exons, cds_range["max"]);
+    std::uint32_t last_idx = get_exon_num(info.exons, info.cds.back().end);
+    std::uint32_t max_idx = get_exon_num(info.exons, cds_max);
     if (max_idx == last_idx) {
-        info.cds.back()[1] = cds_range["max"];
+        info.cds.back().end = cds_max;
     } else {
-        info.cds.back()[1] = info.exons[last_idx][1];  // extend existing CDS
-        std::vector<int> extra_cds = {info.exons[max_idx][0], cds_range["max"]};
-        info.cds.push_back(extra_cds);
+        info.cds.back().end = info.exons[last_idx].end;  // extend existing CDS
+        info.cds.push_back(Region {info.exons[max_idx].start, cds_max});
     }
 }
 
@@ -82,8 +82,8 @@ static void set_span(TxInfo & info) {
     bool first = true;
     for (auto regions : {&info.exons, &info.cds}) {
         for (auto & x : *regions) {
-            info.start = first ? x[0] : std::min(info.start, x[0]);
-            info.end = first ? x[1] : std::max(info.end, x[1]);
+            info.start = first ? x.start : std::min(info.start, x.start);
+            info.end = first ? x.end : std::max(info.end, x.end);
             first = false;
         }
     }
@@ -103,9 +103,9 @@ static void add_transcript(std::vector<NamedTx> & transcripts, TxInfo & info,
         set_span(info);
         Tx tx = Tx(info.name, info.chrom, info.start, info.end, info.strand[0],
             info.transcript_type, info.attributes);
-        tx.set_exons(info.exons);
-        tx.set_cds(info.cds);
-        transcripts.push_back({symbol, alt_ids, tx, info.is_canonical});
+        tx.set_exons(std::move(info.exons));
+        tx.set_cds(std::move(info.cds));
+        transcripts.push_back({std::move(symbol), std::move(alt_ids), std::move(tx), info.is_canonical});
     } catch (const std::invalid_argument & e) {
         std::cerr << "skipping transcript " << info.name << ": " << e.what() << std::endl;
     }
@@ -200,11 +200,11 @@ static void load_transcripts(std::vector<NamedTx> & transcripts, GTF &gtf_file,
             info.end = gtf.end;
             info.attributes = std::move(gtf.attributes);
         } else if (gtf.feature == "CDS") {
-            info.cds.push_back(std::vector<int> {gtf.start, gtf.end});
+            info.cds.push_back(Region {gtf.start, gtf.end});
             cds_range["max"] = std::max(std::max(cds_range["max"], gtf.start), gtf.end);
             cds_range["min"] = std::min(std::min(cds_range["min"], gtf.start), gtf.end);
         } else if (gtf.feature == "exon") {
-            info.exons.push_back(std::vector<int> {gtf.start, gtf.end});
+            info.exons.push_back(Region {gtf.start, gtf.end});
         } else if ((gtf.feature == "stop_codon") || (gtf.feature == "start_codon")) {
             cds_range["max"] = std::max(std::max(cds_range["max"], gtf.start), gtf.end);
             cds_range["min"] = std::min(std::min(cds_range["min"], gtf.start), gtf.end);

{gencodegenes-1.2.0 → gencodegenes-1.2.1}/src/gencode.h

@@ -21,8 +21,8 @@ struct TxInfo {
     int end = 0;
     std::string strand;
     std::string transcript_type;
-    std::vector<std::vector<int> > exons;
-    std::vector<std::vector<int> > cds;
+    std::vector<Region> exons;
+    std::vector<Region> cds;
     int offset = 0;
     int is_canonical = 0;
     std::map<std::string, std::string> attributes;