gencodegenes 1.1.6__tar.gz → 1.2.0__tar.gz

{gencodegenes-1.1.6 → gencodegenes-1.2.0}/MANIFEST.in

@@ -5,10 +5,10 @@ include src/gencodegenes/*.cpp
 include src/gencodegenes/*.py
 include src/gencodegenes/*.pyx
 include src/gencodegenes/*.pxd
+include src/gencodegenes/*.pyi
+include src/gencodegenes/py.typed
 include src/*.h
 include src/*.cpp
-include src/gzstream/gzstream.C
-include src/gzstream/gzstream.h
 
 include src/zlib/*.c
 include src/zlib/*.h

{gencodegenes-1.1.6/src/gencodegenes.egg-info → gencodegenes-1.2.0}/PKG-INFO

@@ -1,13 +1,13 @@
 Metadata-Version: 2.4
 Name: gencodegenes
-Version: 1.1.6
+Version: 1.2.0
 Summary: Package to load genes from GENCODE GTF files
 Author-email: Jeremy McRae <jeremy.mcrae@gmail.com>
+License-Expression: MIT
 Project-URL: homepage, https://github.com/jeremymcrae/gencodegenes
-Classifier: License :: OSI Approved :: MIT License
 Classifier: Development Status :: 4 - Beta
 Classifier: Topic :: Scientific/Engineering :: Bio-Informatics
-Requires-Python: >=3.8
+Requires-Python: >=3.10
 Description-Content-Type: text/markdown
 License-File: LICENSE.txt
 Requires-Dist: pyfaidx>=0.5.8
@@ -16,7 +16,7 @@ Dynamic: license-file
 
 ### GENCODEGenes
 
-This package loads genes from GENCODE GTF/GFF files, groups transcripts by gene, 
+This package loads genes from GENCODE GTF files, groups transcripts by gene, 
 and provides methods for transcripts, so you can find exon coordinates, CDS 
 distances and sequences.
 
@@ -31,22 +31,30 @@ pip install gencodegenes
 from gencodegenes import Gencode
 
 gencode = Gencode(GTF_PATH)
-# full function arguments are Gencode(gtf_path, fasta_path=None, coding_only=True)
-#  - fasta_path: pass in path to fasta file to get gene transcripts with sequence
+# full function arguments are Gencode(gencode, fasta=None, coding_only=True)
+#  - gencode: path to GTF file (plain or gzipped)
+#  - fasta: pass in path to fasta file to get gene transcripts with sequence
 #  - coding_only: pass in False to include all transcripts, not just protein coding
 
+# use as a context manager to close the fasta when done
+with Gencode(GTF_PATH, fasta=FASTA_PATH) as gencode:
+    ...
+
 # get gene by HGNC symbol
 gene = gencode['OR5A1']
 transcripts = gene.transcripts
-canonical = gene.canonical  # picks MANE transcript if available, if none named
-                            # as MANE, picks the one tagged as appris_principal
-                            # (or longest CDS if multiple), if none tagged, picks
-                            # the longest protein coding, if none protein coding,
-                            # picks the longest cDNA 
+canonical = gene.canonical  # picks the transcript tagged Ensembl_canonical (the
+                            # MANE Select transcript in human, where one exists), if
+                            # none tagged, picks from those tagged appris_principal,
+                            # if none tagged, picks from all transcripts. Within
+                            # these, picks the longest CDS, or the longest cDNA if
+                            # none are protein coding
 gene.start, gene.end, gene.chrom, gene.strand, gene.symbol # other attributes available
+gene.alternate_ids  # gene_id and hgnc_id from the GTF
 
 
-# find gene nearest a genomic position, or overlapping a genomic region
+# find gene nearest a genomic position, or overlapping a genomic region. 
+# Chromosomes match with or without the 'chr' prefix
 gencode.nearest('chr1', 1000000)
 gencode.in_region('chr1', 1000000, 2000000)
 
@@ -58,14 +66,15 @@ tx.get_coding_distance(pos)              # get distance in CDS to CDS start
 tx.get_closest_exon(pos)                 # find exon closest to position
 tx.get_position_on_chrom(cds_pos)        # convert CDS pos to genomic pos
 tx.get_codon_info(pos)                   # get info about codon for a site
-tx.get_codon_number_for_cds_pos(cds_pos) # convert CDS pos to codon number
-tx.translate(seq)                        # translate DNA to AA (if opened with Fasta)
+tx.get_codon_number_for_cds_position(cds_pos) # convert CDS pos to codon number
+tx.translate(seq)                        # translate DNA to AA
+tx.consequence(pos, ref, alt)            # get variant consequence (if opened with fasta)
 
 # the transcript also has associated data fields
 tx.name         # transcript ID
 tx.chrom        # transcript chromosome
-tx.start        # transcript start (TSS)
-tx.end          # transcript end
+tx.start        # transcript start (lowest position on chromosome)
+tx.end          # transcript end (highest position on chromosome)
 tx.cds_start    # CDS start position
 tx.cds_end      # CDS end position 
 tx.type         # transcript type e.g. protein_coding
@@ -73,5 +82,6 @@ tx.strand       # strand (+ or -)
 tx.exons        # list of exon coordinates
 tx.cds          # list of CDS coordinates
 tx.cds_sequence # get cDNA sequence (if Gencode was opened with fasta)
+tx.attributes   # dict-like view of the GTF attributes e.g. tx.attributes['gene_id']
 
 ```

{gencodegenes-1.1.6 → gencodegenes-1.2.0}/README.md

@@ -1,7 +1,7 @@
 
 ### GENCODEGenes
 
-This package loads genes from GENCODE GTF/GFF files, groups transcripts by gene, 
+This package loads genes from GENCODE GTF files, groups transcripts by gene, 
 and provides methods for transcripts, so you can find exon coordinates, CDS 
 distances and sequences.
 
@@ -16,22 +16,30 @@ pip install gencodegenes
 from gencodegenes import Gencode
 
 gencode = Gencode(GTF_PATH)
-# full function arguments are Gencode(gtf_path, fasta_path=None, coding_only=True)
-#  - fasta_path: pass in path to fasta file to get gene transcripts with sequence
+# full function arguments are Gencode(gencode, fasta=None, coding_only=True)
+#  - gencode: path to GTF file (plain or gzipped)
+#  - fasta: pass in path to fasta file to get gene transcripts with sequence
 #  - coding_only: pass in False to include all transcripts, not just protein coding
 
+# use as a context manager to close the fasta when done
+with Gencode(GTF_PATH, fasta=FASTA_PATH) as gencode:
+    ...
+
 # get gene by HGNC symbol
 gene = gencode['OR5A1']
 transcripts = gene.transcripts
-canonical = gene.canonical  # picks MANE transcript if available, if none named
-                            # as MANE, picks the one tagged as appris_principal
-                            # (or longest CDS if multiple), if none tagged, picks
-                            # the longest protein coding, if none protein coding,
-                            # picks the longest cDNA 
+canonical = gene.canonical  # picks the transcript tagged Ensembl_canonical (the
+                            # MANE Select transcript in human, where one exists), if
+                            # none tagged, picks from those tagged appris_principal,
+                            # if none tagged, picks from all transcripts. Within
+                            # these, picks the longest CDS, or the longest cDNA if
+                            # none are protein coding
 gene.start, gene.end, gene.chrom, gene.strand, gene.symbol # other attributes available
+gene.alternate_ids  # gene_id and hgnc_id from the GTF
 
 
-# find gene nearest a genomic position, or overlapping a genomic region
+# find gene nearest a genomic position, or overlapping a genomic region. 
+# Chromosomes match with or without the 'chr' prefix
 gencode.nearest('chr1', 1000000)
 gencode.in_region('chr1', 1000000, 2000000)
 
@@ -43,14 +51,15 @@ tx.get_coding_distance(pos)              # get distance in CDS to CDS start
 tx.get_closest_exon(pos)                 # find exon closest to position
 tx.get_position_on_chrom(cds_pos)        # convert CDS pos to genomic pos
 tx.get_codon_info(pos)                   # get info about codon for a site
-tx.get_codon_number_for_cds_pos(cds_pos) # convert CDS pos to codon number
-tx.translate(seq)                        # translate DNA to AA (if opened with Fasta)
+tx.get_codon_number_for_cds_position(cds_pos) # convert CDS pos to codon number
+tx.translate(seq)                        # translate DNA to AA
+tx.consequence(pos, ref, alt)            # get variant consequence (if opened with fasta)
 
 # the transcript also has associated data fields
 tx.name         # transcript ID
 tx.chrom        # transcript chromosome
-tx.start        # transcript start (TSS)
-tx.end          # transcript end
+tx.start        # transcript start (lowest position on chromosome)
+tx.end          # transcript end (highest position on chromosome)
 tx.cds_start    # CDS start position
 tx.cds_end      # CDS end position 
 tx.type         # transcript type e.g. protein_coding
@@ -58,5 +67,6 @@ tx.strand       # strand (+ or -)
 tx.exons        # list of exon coordinates
 tx.cds          # list of CDS coordinates
 tx.cds_sequence # get cDNA sequence (if Gencode was opened with fasta)
+tx.attributes   # dict-like view of the GTF attributes e.g. tx.attributes['gene_id']
 
 ```

{gencodegenes-1.1.6 → gencodegenes-1.2.0}/pyproject.toml

@@ -1,13 +1,14 @@
 [build-system]
-requires = ["cython", "setuptools >= 40.6.0", "wheel"]
+requires = ["cython", "setuptools >= 77.0.0"]
 build-backend = "setuptools.build_meta"
 
 [project]
 name = 'gencodegenes'
-version = '1.1.6'
+version = '1.2.0'
 description = 'Package to load genes from GENCODE GTF files'
 readme = 'README.md'
-requires-python = ">=3.8"
+license = "MIT"
+requires-python = ">=3.10"
 authors = [
     {name = 'Jeremy McRae', email = 'jeremy.mcrae@gmail.com'}
 ]
@@ -17,7 +18,6 @@ dependencies = [
 ]
 
 classifiers = [
-    "License :: OSI Approved :: MIT License",
     "Development Status :: 4 - Beta",
     "Topic :: Scientific/Engineering :: Bio-Informatics",
 ]
@@ -27,5 +27,8 @@ packages = [
     "gencodegenes"
 ]
 
+[tool.setuptools.package-data]
+gencodegenes = ['transcript.pxd', 'tx.h', 'tx.cpp', 'py.typed', '*.pyi']
+
 [project.urls]
 homepage = 'https://github.com/jeremymcrae/gencodegenes'

gencodegenes-1.2.0/setup.py

@@ -0,0 +1,69 @@
+
+import glob
+import os
+import sys
+import shutil
+
+from setuptools import setup, Extension
+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",
+            ]
+
+gencode_sources = [
+    "src/gencodegenes/gencode.pyx",
+    "src/gencode.cpp",
+    "src/gtf.cpp",
+    "src/tx.cpp",
+]
+
+libs = ['z']
+include_dirs = ['src/']
+
+if sys.platform == 'win32':
+    gencode_sources += glob.glob('src/zlib/*.c')
+    include_dirs.append('src/zlib/')
+    libs = []
+
+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=gencode_sources,
+        include_dirs=include_dirs,
+        libraries=libs,
+        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),
+    )

{gencodegenes-1.1.6 → gencodegenes-1.2.0}/src/gencode.cpp

@@ -1,10 +1,12 @@
 
 #include <algorithm>
 #include <cstdint>
+#include <deque>
 #include <string>
 #include <map>
 #include <set>
 #include <stdexcept>
+#include <unordered_map>
 #include <vector>
 
 #include <iostream>
@@ -72,71 +74,131 @@ static void include_end_codons(std::map<std::string, int> cds_range, TxInfo & in
     }
 }
 
+// set the transcript span from its exons and CDS, for GTFs without transcript lines
+static void set_span(TxInfo & info) {
+    if (info.start != 0 || info.end != 0) {
+        return;
+    }
+    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]);
+            first = false;
+        }
+    }
+}
+
+// construct a Tx from the features collected for a transcript
+//
+// Transcripts with inconsistent coordinates (e.g. a stop codon outside the
+// exons) are skipped with a warning, so one malformed transcript doesn't stop
+// the rest of the GTF from loading.
+static void add_transcript(std::vector<NamedTx> & transcripts, TxInfo & info,
+        std::map<std::string, int> & cds_range, std::string & symbol,
+        std::vector<std::string> & alt_ids) {
+    try {
+        // adjust CDS for start and stop codon coords
+        include_end_codons(cds_range, 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});
+    } catch (const std::invalid_argument & e) {
+        std::cerr << "skipping transcript " << info.name << ": " << e.what() << std::endl;
+    }
+}
+
+// features collected for a transcript, while loading the GTF
+struct TxEntry {
+    TxInfo info;
+    std::map<std::string, int> cds_range = {{"max", 0}, {"min", 999999999}};
+    std::string symbol;
+    std::vector<std::string> alt_ids;
+};
+
+// build transcripts in the order they first appeared, freeing each entry once
+// its transcript is built
+//
+// @param pos start of the current GTF line. Transcripts are only built once this
+//     passes their end, as no further lines for them can follow in a
+//     position-sorted GTF. Use -1 to build all transcripts.
+static void build_transcripts(std::vector<NamedTx> & transcripts,
+        std::deque<TxEntry> & entries, std::unordered_map<std::string, TxEntry *> & index,
+        int pos=-1) {
+    while (!entries.empty()) {
+        TxEntry & x = entries.front();
+        if (pos != -1 && (x.info.end == 0 || pos <= x.info.end)) {
+            break;
+        }
+        index.erase(x.info.name);
+        add_transcript(transcripts, x.info, x.cds_range, x.symbol, x.alt_ids);
+        entries.pop_front();
+    }
+}
+
 // collect all features for a transcript into a single object
 //
 // When we load lines from gencode GTF files, each line represents a single exon
-// or CDS, and we need to combine these based on transcript ID
+// or CDS, and we need to combine these based on transcript ID. Lines for a
+// transcript are usually contiguous, but position-sorted GTFs interleave
+// transcripts, so features are collected by transcript ID until the chromosome
+// changes (GTFs are grouped by chromosome), or the GTF moves past the transcript.
 static void load_transcripts(std::vector<NamedTx> & transcripts, GTF &gtf_file, bool coding=true) {
     std::set<std::string> permit = {"exon", "CDS", "UTR", "transcript", 
         "stop_codon", "start_codon"};
-    std::map<std::string, int> cds_range = {{"max", 0}, {"min", 999999999}};
-    std::string tx_id = "";
-    std::string symbol = "";
-    std::vector<std::string> alt_ids;
-    std::string current;
-    TxInfo info;
+    std::deque<TxEntry> entries;
+    std::unordered_map<std::string, TxEntry *> index;
+    TxEntry * entry = nullptr;
 
     GTFLine gtf;
 
-    while (true) {
-        try {
-            gtf = gtf_file.next();
-        } catch (const std::out_of_range& e) {
-            break;
-        }
-        
-
+    while (gtf_file.next(gtf)) {
         if (permit.count(gtf.feature) == 0) {
             continue;
         } else if (coding && (gtf.transcript_type != "protein_coding")) {
             continue;
         }
 
-        current = gtf.tx_id;
-        if (tx_id == "") {
-            tx_id = current;
-            symbol = gtf.symbol;
-            alt_ids = gtf.alternate_ids;
-        }
-
-        if (tx_id != current) {
-            // adjust CDS for start and stop codon coords
-            include_end_codons(cds_range, info);
-            Tx tx = Tx(info.name, info.chrom, info.start, info.end, info.strand[0], 
-                info.transcript_type);
-            tx.set_exons(info.exons);
-            tx.set_cds(info.cds);
-            transcripts.push_back({symbol, alt_ids, tx, info.is_canonical});
-            info = {};
-            tx_id = current;
-            cds_range["max"] = 0;
-            cds_range["min"] = 999999999;
-            symbol = gtf.symbol;
-            alt_ids = gtf.alternate_ids;
-            info.is_canonical = false;
-        }
-
-        if (info.name == "") {
-            info.name = tx_id;
-            info.chrom = gtf.chrom;
-            info.strand = gtf.strand;
-            info.is_canonical = gtf.is_canonical;
-            info.transcript_type = gtf.transcript_type;
+        // only look up the transcript when it differs from the previous line's
+        if (entry == nullptr || gtf.tx_id != entry->info.name || gtf.chrom != entry->info.chrom) {
+            int pos = (entry != nullptr && gtf.chrom != entry->info.chrom) ? -1 : gtf.start;
+            build_transcripts(transcripts, entries, index, pos);
+            auto it = index.find(gtf.tx_id);
+            if (it != index.end()) {
+                entry = it->second;
+            } else {
+                entries.emplace_back();
+                entry = &entries.back();
+                index[gtf.tx_id] = entry;
+                
+                TxInfo & info = entry->info;
+                info.name = gtf.tx_id;
+                info.chrom = gtf.chrom;
+                info.strand = gtf.strand;
+                info.is_canonical = gtf.is_canonical;
+                info.transcript_type = gtf.transcript_type;
+                entry->symbol = gtf.symbol;
+                entry->alt_ids = gtf.alternate_ids;
+                if (gtf.feature != "transcript") {
+                    // without a transcript line, use the first line's attributes,
+                    // minus the fields specific to that feature
+                    info.attributes = gtf.attributes;
+                    for (auto field : {"exon_number", "exon_id", "exon_version"}) {
+                        info.attributes.erase(field);
+                    }
+                }
+            }
         }
 
+        TxInfo & info = entry->info;
+        std::map<std::string, int> & cds_range = entry->cds_range;
         if (gtf.feature == "transcript") {
             info.start = gtf.start;
             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});
             cds_range["max"] = std::max(std::max(cds_range["max"], gtf.start), gtf.end);
@@ -149,14 +211,7 @@ static void load_transcripts(std::vector<NamedTx> & transcripts, GTF &gtf_file,
         }
     }
 
-    // also include the final transcript (if it transcript exists)
-    if (info.name != "") {
-        include_end_codons(cds_range, info);
-        Tx tx = Tx(info.name, info.chrom, info.start, info.end, info.strand[0], info.transcript_type);
-        tx.set_exons(info.exons);
-        tx.set_cds(info.cds);
-        transcripts.push_back({symbol, alt_ids, tx, info.is_canonical});
-    }
+    build_transcripts(transcripts, entries, index);
 }
 
 std::vector<NamedTx> open_gencode(std::string path, bool coding) {
@@ -176,10 +231,6 @@ std::vector<std::string> _in_region(std::string chrom, int start, int end,
         std::map<std::string, std::vector<GenePoint>> & ends,
         int max_window=2500000) {
     
-    if (chrom.size() < 3 || chrom.substr(0, 3) != "chr") {
-        chrom.insert(0, "chr");
-    }
-    
     if (starts.count(chrom) == 0) {
         throw std::invalid_argument("unknown_chrom: " + chrom);
     }
@@ -256,7 +307,7 @@ std::vector<std::string> _in_region(std::string chrom, int start, int end,
 //     gencode::open_gencode(path);
 // }
 // 
-// g++ -std=c++11 gencode.cpp gtf.cpp tx.cpp gzstream/gzstream.C -Igzstream -lz
+// g++ -std=c++11 gencode.cpp gtf.cpp tx.cpp -lz
 
 
 

{gencodegenes-1.1.6 → gencodegenes-1.2.0}/src/gencode.h

@@ -17,14 +17,15 @@ namespace gencode {
 struct TxInfo {
     std::string name = "";
     std::string chrom;
-    int start;
-    int end;
+    int start = 0;
+    int end = 0;
     std::string strand;
     std::string transcript_type;
     std::vector<std::vector<int> > exons;
     std::vector<std::vector<int> > cds;
     int offset = 0;
     int is_canonical = 0;
+    std::map<std::string, std::string> attributes;
 };
 
 // stores HGNC symbol with the transcript, so we can collect transcripts by gene