Merge pull request #19 from michael-kotliar/master

Export 1-based coordinates

Author: nsalomonis

altanalyze3/components/aggregate/main.py

@@ -53,15 +53,25 @@ def __init__(self):
     def add(self, sa, ea):                                                                                           # sa - start annotation, ea - end annotation
 
         sa_shift = "" if sa.position == 0 else f"""_{sa.position}"""                                                 # junction start coord if not exact exon match
-        ea_shift = "" if ea.position == 0 else f"""_{ea.position}"""                                                 # junction end coorf if not exact exon match
+        ea_shift = "" if ea.position == 0 else f"""_{ea.position + 1}"""                                             # junction end coorf if not exact exon match
 
         if sa.match == AnnMatchCat.CLOSEST or ea.match == AnnMatchCat.CLOSEST:                                       # not enough data to make a conclusion
             pass
         elif sa.gene == ea.gene and sa.strand == ea.strand:                                                          # the same gene and strand
             if sa.match == AnnMatchCat.EXON_END and ea.match == AnnMatchCat.EXON_START:                              #    exact match
-                self.__exact_match.append((f"""{sa.gene}:{sa.exon}-{ea.exon}""", sa.strand))
+                if sa.strand == "+":
+                    self.__exact_match.append((f"""{sa.gene}:{sa.exon}-{ea.exon}""", sa.strand))                     #        start - end exons order
+                elif sa.strand == "-":
+                    self.__exact_match.append((f"""{sa.gene}:{ea.exon}-{sa.exon}""", sa.strand))                     #        end - start exons order
+                else:
+                    assert(False), "Not implemented logic"
             else:                                                                                                    #    partial match
-                self.__partial_match.append((f"""{sa.gene}:{sa.exon}{sa_shift}-{ea.exon}{ea_shift}""", sa.strand))
+                if sa.strand == "+":
+                    self.__partial_match.append((f"""{sa.gene}:{sa.exon}{sa_shift}-{ea.exon}{ea_shift}""", sa.strand))  #     start - end exons order
+                elif sa.strand == "-":
+                    self.__partial_match.append((f"""{sa.gene}:{ea.exon}{ea_shift}-{sa.exon}{sa_shift}""", sa.strand))  #     end - start exons order
+                else:
+                    assert(False), "Not implemented logic"
         else:                                                                                                        # different gene and/or strand
             self.__distant_match.append((f"""{sa.gene}:{sa.exon}{sa_shift}-{ea.gene}:{ea.exon}{ea_shift}""", "."))   #    distant match
 

altanalyze3/utilities/io.py

@@ -140,9 +140,9 @@ def export_counts_to_anndata(counts_df, location, counts_columns=None, metadata_
 
     def __get_name(series, strand_coords):
         if strand_coords and "strand" in series and series.at["strand"] == "-":
-            return f"""{series.at["chr"]}:{series.at["end"]}-{series.at["start"]}"""
+            return f"""{series.at["chr"]}:{series.at["end"] + 1}-{series.at["start"]}"""
         else:
-            return f"""{series.at["chr"]}:{series.at["start"]}-{series.at["end"]}"""
+            return f"""{series.at["chr"]}:{series.at["start"]}-{series.at["end"] + 1}"""
 
     csr_matrix = counts_df.loc[:, counts_columns].astype(pandas.SparseDtype(sparse_dtype, fill_value)).T.sparse.to_coo().tocsr()
     adata = anndata.AnnData(csr_matrix, dtype=sparse_dtype)

altanalyze3/utilities/parser.py

@@ -223,7 +223,7 @@ def get_parser(self):
         )
         aggregate_parser.add_argument(
             "--bed",
-            help="Export annotated coordinates as BED file. Default: False",
+            help="Export annotated 0-based coordinates as BED file. Default: False",
             action="store_true"
         )
         self.add_common_arguments(aggregate_parser)

tests/unit_tests/test_data.py

@@ -13,7 +13,7 @@
     "Cal27P5-3.bam": "3d6fdef5e90c16deed6c36b418c597c4",
     "Cal27P5-3.bam.bai": "c92fbf32ffcf874979c41051f0d2e16c",
     "Cal27P5-1-copy.bam": "6e418cb604ea82c90d4b3e832afb4997",
-    "Cal27P5_1_aggregated_counts.bed.gz": "26251bd42bac659434799040c8b0caf8",
+    "Cal27P5_1_aggregated_counts.bed.gz": "faee5946f7d18064773fc67e01e4ce5b",
     "Cal27P5_1_intcounts.bed": "6f5250f1a841438fadcbe1419966f133",
     "Cal27P5_1_juncounts.bed": "24f6a6ac5f831c7709102c235db03ec5",
     "Cal27P5_2_intcounts.bed": "2d9889a6520f96012ba7d4be5cdcfea3",