fixing off-by one error at transcript start, problematic intergenic v…

…ariants, and only one assembly mapper on pretty print.

src/hgvs/easy.py

@@ -67,7 +67,8 @@
 t_to_g = projector.t_to_g
 t_to_p = projector.t_to_p
 get_relevant_transcripts = am38.relevant_transcripts
-pretty = PrettyPrint(hdp, useColor=True, showLegend=True)
+pretty37 = PrettyPrint(hdp, am37, useColor=True, showLegend=True)
+pretty38 = PrettyPrint(hdp, am38, useColor=True, showLegend=True)
 
 # <LICENSE>
 # Copyright 2018 HGVS Contributors (https://github.com/biocommons/hgvs)

src/hgvs/pretty/console/tx_alig_renderer.py

@@ -35,7 +35,7 @@ def display(self, data: VariantData) -> str:
 
             counter += 1
 
-            if not pdata.mapped_pos:
+            if pdata.mapped_pos is None:
                 var_str += " "
                 continue
 

src/hgvs/pretty/console/tx_mapping_renderer.py

@@ -40,12 +40,12 @@ def display(self, data: VariantData) -> str:
                 prev_c_pos = c_pos
                 continue
 
-            if (c_pos + 1) % 10 == 0:
+            if (c_pos) % 10 == 0:
                 var_str += "|"
                 prev_c_pos = c_pos
                 continue
 
-            elif (c_pos + 1) % 5 == 0:
+            elif (c_pos) % 5 == 0:
                 var_str += "."
                 prev_c_pos = c_pos
                 continue
@@ -69,7 +69,7 @@ def display(self, data: VariantData) -> str:
                 prev_c_pos = c_pos
                 continue
 
-            elif c_pos == 0:
+            elif c_pos == 1:
                 var_str += "|"
 
             var_str += " "

src/hgvs/pretty/console/tx_pos.py

@@ -33,13 +33,13 @@ def display(self, data: VariantData) -> str:
             if len(var_str) > count:
                 continue
 
-            if (c_pos + 1) % 10 == 0:
+            if (c_pos) % 10 == 0:
                 # if pdata.c_interval.start.datum == Datum.CDS_END:
                 #     var_str += "*"
                 var_str += f"{interval} "
                 continue
 
-            elif c_pos == 0:
+            elif c_pos == 1:
                 var_str += f"{interval} "
                 continue
             var_str += " "

src/hgvs/pretty/datacompiler.py

@@ -206,24 +206,19 @@ def data(
 
         if tx_ac:
             tx_seq = self.config.hdp.get_seq(tx_ac)
-            if self.config.default_assembly == "GRCh37":
-                am = self.config.am37
-            else:
-                am = self.config.am38
-
-            mapper = am._fetch_AlignmentMapper(
+
+            mapper = self.config.assembly_mapper._fetch_AlignmentMapper(
                 tx_ac=tx_ac, alt_ac=var_g.ac, alt_aln_method="splign"
             )
 
         else:
             tx_seq = ""
             mapper = None
-        # print(tx_seq)
 
         # we don't know the protein ac, get it looked up:
         pro_ac = None
         if var_c_or_n and var_c_or_n.type == "c":
-            var_p = am.c_to_p(var_c_or_n)
+            var_p = self.config.assembly_mapper.c_to_p(var_c_or_n)
             reference_data = RefTranscriptData(self.config.hdp, tx_ac, pro_ac)
         else:
             var_p = None
@@ -277,8 +272,8 @@ def data(
                         reference_data,
                         pdata,
                         cig,
-                        prev_c_pos + 1,
-                        prev_n_pos + 1,
+                        prev_c_pos,
+                        prev_n_pos,
                     )
 
                     exon_nr, feat = self._get_exon_nr(tx_exons, chromosome_pos)
@@ -292,7 +287,7 @@ def data(
                     pdata.mapped_pos = prev_mapped_pos
                     pdata.mapped_pos_offset = mapped_pos_offset
                     pdata.cigar_ref = cig
-                    pdata.ref = chrom_seq[chromosome_pos]
+                    pdata.ref = chrom_seq[chromosome_pos-1]
 
                     if mapper.strand > 0:
                         prev_c_pos += 1
@@ -321,12 +316,12 @@ def data(
             pdata.n_interval = n_interval
             if c_interval is not None:
                 pdata.c_interval = c_interval
-                c_pos = int(c_interval.start.base) - 1
+                c_pos = int(c_interval.start.base)
                 prev_c_pos = c_pos
             else:
                 prev_c_pos = -1
 
-            n_pos = int(n_interval.start.base) - 1
+            n_pos = int(n_interval.start.base)
             prev_n_pos = n_pos
 
             self._populate_with_n_c(
@@ -351,7 +346,7 @@ def data(
             pd = reversed(position_details)
             position_details = list(pd)
 
-        is_rna = var_c_or_n.ac.startswith("NR_") # not sure how to check this for ENSTs
+        is_rna = var_c_or_n and var_c_or_n.ac.startswith("NR_") # not sure how to check this for ENSTs
 
         vd = VariantData(
             seq_start,
@@ -421,10 +416,10 @@ def _populate_with_n_c(
         n_interval,
         c_interval,
     ):
-        n_pos = int(n_interval.start.base) - 1
+        n_pos = int(n_interval.start.base) 
 
         if c_interval:
-            c_pos = int(c_interval.start.base) - 1
+            c_pos = int(c_interval.start.base) 
             pdata.c_pos = c_pos
             pdata.c_offset = c_interval.start.offset
         else:
@@ -435,7 +430,7 @@ def _populate_with_n_c(
 
         pdata.n_pos = n_pos
 
-        pdata.tx = tx_seq[pdata.n_pos]
+        pdata.tx = tx_seq[pdata.n_pos-1]
 
         coding = True
         if var_c_or_n.type == "n":  # rna coding can't be in protein space

src/hgvs/pretty/models.py

@@ -98,11 +98,9 @@ class PrettyConfig:
     """A container for various configurations."""
 
     hdp: Interface
-    am37: AssemblyMapper
-    am38: AssemblyMapper
+    assembly_mapper: AssemblyMapper
     padding_left: int = 20
     padding_right: int = 20
-    default_assembly: str = "GRCh37"
     useColor: bool = False
     showLegend: bool = True
     showAllShuffleableRegions = False

src/hgvs/pretty/pretty_print.py

@@ -27,7 +27,7 @@ class PrettyPrint:
     def __init__(
         self,
         hdp: hgvs.dataproviders.interface.Interface,
-        default_assembly: str = "GRCh37",
+        assembly_mapper:AssemblyMapper,
         padding_left: int = 20,
         padding_right: int = 20,
         useColor=False,
@@ -44,37 +44,24 @@ def __init__(
 
         :param padding: spacing left and right of the variant for display purposes.
         """
-        am37: AssemblyMapper = AssemblyMapper(hdp, assembly_name="GRCh37", alt_aln_method=alt_aln_method)
-        am38: AssemblyMapper = AssemblyMapper(hdp, assembly_name="GRCh38", alt_aln_method=alt_aln_method)
-
+
         self.config = PrettyConfig(
-            hdp,
-            am37,
-            am38,
-            padding_left,
-            padding_right,
-            default_assembly,
-            useColor,
-            showLegend,
-            infer_hgvs_c,
-            all,
-            show_reverse_strand,
-            alt_aln_method,
-            reverse_display
+            hdp=hdp,
+            assembly_mapper=assembly_mapper,
+            padding_left=padding_left,
+            padding_right=padding_right,
+            useColor=useColor,
+            showLegend=showLegend,
+            infer_hgvs_c=infer_hgvs_c,
+            all=all,
+            show_reverse_strand=show_reverse_strand,
+            alt_aln_method=alt_aln_method,
+            reverse_display=reverse_display
         )
 
-    def _get_assembly_mapper(self) -> AssemblyMapper:
-        if self.config.default_assembly == "GRCh37":
-            am = self.config.am37
-        else:
-            am = self.config.am38
-
-        return am
-
     def _get_all_transcripts(self, var_g) -> List[str]:
-        am = self._get_assembly_mapper()
-
-        transcripts = am.relevant_transcripts(var_g)
+
+        transcripts = self.config.assembly_mapper.relevant_transcripts(var_g)
 
         return transcripts
 
@@ -87,7 +74,7 @@ def _infer_hgvs_c(self, var_g: SequenceVariant, tx_ac: str = None) -> SequenceVa
             else:
                 return None
 
-        am = self._get_assembly_mapper()
+        am = self.config.assembly_mapper
 
         if tx_ac.startswith("NR_"):
             var_n = am.g_to_n(var_g, tx_ac)
@@ -97,10 +84,7 @@ def _infer_hgvs_c(self, var_g: SequenceVariant, tx_ac: str = None) -> SequenceVa
 
     def _map_to_chrom(self, sv: SequenceVariant) -> SequenceVariant:
         """maps a variant to chromosomal coords, if needed."""
-        if self.config.default_assembly == "GRCh37":
-            am = self.config.am37
-        else:
-            am = self.config.am38
+        am = self.config.assembly_mapper
 
         if sv.type == "c":
             return am.c_to_g(sv)

src/hgvs/shell.py

@@ -72,7 +72,8 @@ def shell():
         normalizer,
         parse,
         parser,
-        pretty,
+        pretty37,
+        pretty38,
         projector,
         t_to_g,
         t_to_p,