gencodegenes 1.0.8__tar.gz → 1.0.10__tar.gz

{gencodegenes-1.0.8/src/gencodegenes.egg-info → gencodegenes-1.0.10}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: gencodegenes
-Version: 1.0.8
+Version: 1.0.10
 Summary: Package to load genes from GENCODE GTF files
 Home-page: https://github.com/jeremymcrae/gencodegenes
 Author: Jeremy McRae

{gencodegenes-1.0.8 → gencodegenes-1.0.10}/setup.py

@@ -112,14 +112,13 @@ setup(name="gencodegenes",
     description='Package to load genes from GENCODE GTF files',
     long_description=io.open('README.md', encoding='utf-8').read(),
     long_description_content_type='text/markdown',
-    version="1.0.8",
+    version="1.0.10",
     author="Jeremy McRae",
     author_email="jeremy.mcrae@gmail.com",
     license="MIT",
     url='https://github.com/jeremymcrae/gencodegenes',
     packages=["gencodegenes"],
     install_requires=[
-        'cython >= 0.27.0',
         'pyfaidx >= 0.5.8',
     ],
     package_dir={'': 'src'},

{gencodegenes-1.0.8 → gencodegenes-1.0.10}/src/gencode.cpp

@@ -16,16 +16,16 @@
 namespace gencode {
 
 // check which exon is first, by start position
-bool compareExons(std::vector<int> e1, std::vector<int> e2) {
+static bool compareExons(std::vector<int> e1, std::vector<int> e2) {
     return (e1[0] < e2[0]);
 }
 
-void sort_exons(std::vector<std::vector<int> > & exons) {
+static void sort_exons(std::vector<std::vector<int> > & exons) {
     std::sort(exons.begin(), exons.end(), compareExons);
 }
 
 // find the index of the exon containing a given chromosome position
-std::uint32_t get_exon_num(std::vector<std::vector<int> > exons, int pos) {
+static std::uint32_t get_exon_num(std::vector<std::vector<int> > exons, int pos) {
     for (std::uint32_t i=0; i<exons.size(); i++) {
         if ((pos >= exons[i][0]) && (pos <= exons[i][1])) {
             return i;
@@ -41,7 +41,7 @@ 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.
-void include_end_codons(std::map<std::string, int> cds_range, TxInfo & info) {
+static void include_end_codons(std::map<std::string, int> cds_range, TxInfo & info) {
     if (info.cds.size() == 0) {
         return;
     }
@@ -76,12 +76,13 @@ void include_end_codons(std::map<std::string, int> cds_range, TxInfo & info) {
 //
 // 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
-void load_transcripts(std::vector<NamedTx> & transcripts, GTF &gtf_file, bool coding=true) {
+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;
 
@@ -105,6 +106,7 @@ void load_transcripts(std::vector<NamedTx> & transcripts, GTF &gtf_file, bool co
         if (tx_id == "") {
             tx_id = current;
             symbol = gtf.symbol;
+            alt_ids = gtf.alternate_ids;
         }
 
         if (tx_id != current) {
@@ -114,12 +116,13 @@ void load_transcripts(std::vector<NamedTx> & transcripts, GTF &gtf_file, bool co
                 info.transcript_type);
             tx.set_exons(info.exons);
             tx.set_cds(info.cds);
-            transcripts.push_back({symbol, tx, info.is_canonical});
+            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;
         }
 
@@ -152,7 +155,7 @@ void load_transcripts(std::vector<NamedTx> & transcripts, GTF &gtf_file, bool co
         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, tx, info.is_canonical});
+        transcripts.push_back({symbol, alt_ids, tx, info.is_canonical});
     }
 }
 

{gencodegenes-1.0.8 → gencodegenes-1.0.10}/src/gencode.h

@@ -30,6 +30,7 @@ struct TxInfo {
 // stores HGNC symbol with the transcript, so we can collect transcripts by gene
 struct NamedTx {
     std::string symbol;
+    std::vector<std::string> alternate_ids;
     Tx tx;
     int is_canonical;
 };

gencodegenes-1.0.10/src/gencodegenes/__init__.py

@@ -0,0 +1,7 @@
+from importlib.metadata import version
+
+__name__ = 'gencodegenes'
+__version__ = version('gencodegenes')
+
+from gencodegenes.gencode import Gencode, Gene
+from gencodegenes.transcript import Transcript