code clean

VariantParams better reflects its use. Split apart variants.py as the
code was getting way too long

docs/api/conf.py

@@ -17,7 +17,7 @@
 # -- Project information -----------------------------------------------------
 
 project = 'Truvari'
-copyright = '2023, Adam English'
+copyright = '2025, Adam English'
 author = 'Adam English'
 
 

docs/api/truvari.examples.rst

@@ -32,12 +32,12 @@ The `truvari.VariantRecord` simplifies comparing two VCF entries.
     print("Entries' Size Similarity:", match.sizesim)
     print("Is the match above thresholds:", match.state)
 
-This returns a `truvari.MatchResult`. You can customize matching thresholds by providing a `truvari.Matcher` to the `truvari.VariantFile`.
+This returns a `truvari.MatchResult`. You can customize matching thresholds by providing a `truvari.VariantParams` to the `truvari.VariantFile`.
 
 .. code-block:: python
 
     # Disable sequence and size similarity; enable reciprocal overlap
-    matcher = truvari.Matcher(seqsim=0, sizesim=0, recovl=0.5)
+    matcher = truvari.VariantParams(seqsim=0, sizesim=0, recovl=0.5)
     vcf = truvari.VariantFile("input.vcf.gz", matcher=matcher)
     entry1 = next(vcf)
     entry2 = next(vcf)
@@ -46,11 +46,11 @@ This returns a `truvari.MatchResult`. You can customize matching thresholds by p
 Filtering Variants
 ------------------
 
-The `truvari.Matcher` provides parameters for filtering variants, such as minimum or maximum SV sizes.
+The `truvari.VariantParams` provides parameters for filtering variants, such as minimum or maximum SV sizes.
 
 .. code-block:: python
 
-    matcher = truvari.Matcher(sizemin=200, sizemax=500)
+    matcher = truvari.VariantParams(sizemin=200, sizemax=500)
     vcf = truvari.VariantFile("input.vcf.gz", matcher=matcher)
     # Retrieve all variant records within sizemin and sizemax
     results = [entry for entry in vcf if not entry.size_filter()]
@@ -65,7 +65,8 @@ To subset a VCF to regions specified in a BED file, use:
 .. code-block:: python
 
     for entry in vcf.bed_fetch("regions.bed"):
-        print(entry.var_type(), entry.size())
+        print("Entry's variant type:", entry.var_type())
+        print("Entry's variant size:", entry.var_size())
 
 If your regions of interest are stored in an in-memory object instead of a BED file, use the `.regions_fetch` method:
 

docs/api/truvari.package.rst

@@ -19,6 +19,9 @@ Variant Handling
 .. autoclass:: VariantRecord
    :members:
 
+.. autoclass:: VariantParams
+   :members:
+
 Extra Methods
 -------------
 .. autofunction:: bed_ranges
@@ -114,9 +117,6 @@ Objects
 .. autoclass:: MatchResult
    :members:
 
-.. autoclass:: Matcher
-   :members:
-
 .. autoclass:: SV
    :members:
 

repo_utils/run_doctests.py

@@ -13,7 +13,8 @@
     comparisons,
     msatovcf,
     utils,
-    variants,
+    variant_file,
+    variant_record,
     vcf2df,
 )
 
@@ -27,7 +28,8 @@ def tester(module):
     return ret.failed
 
 fails = 0
-fails += tester(variants)
+fails += tester(variant_file)
+fails += tester(variant_record)
 fails += tester(comparisons)
 fails += tester(utils)
 fails += tester(vcf2df)

repo_utils/run_unittest.py

@@ -75,8 +75,8 @@
 """
 Filtering logic
 """
-matcher = truvari.Matcher(sizemin=0, sizefilt=0, passonly=True)
-vcf = truvari.VariantFile("repo_utils/test_files/variants/filter.vcf", matcher=matcher)
+p = truvari.VariantParams(sizemin=0, sizefilt=0, passonly=True)
+vcf = truvari.VariantFile("repo_utils/test_files/variants/filter.vcf", params=p)
 for entry in vcf:
     try:
         assert entry.filter_call(), f"Didn't filter {str(entry)}"

truvari/__init__.py

@@ -10,11 +10,11 @@
 :class:`GT`
 :class:`LogFileStderr`
 :class:`MatchResult`
-:class:`Matcher`
 :class:`StatsBox`
 :class:`SV`
 :class:`VariantFile`
 :class:`VariantRecord`
+:class:`VariantParams`
 
 Extra methods:
 
@@ -89,7 +89,6 @@
 
 from truvari.matching import (
     MatchResult,
-    Matcher,
     chunker,
     file_zipper
 )
@@ -132,10 +131,9 @@
     vcf_ranges,
 )
 
-from truvari.variants import (
-    VariantFile,
-    VariantRecord,
-)
+from truvari.variant_file import VariantFile
+from truvari.variant_params import VariantParams
+from truvari.variant_record import VariantRecord
 
 from truvari.vcf2df import (
     GT,

truvari/annotations/bpovl.py

@@ -75,7 +75,7 @@ def _transform():
             if span > args.spanmax:
                 continue
 
-            if entry.size() < args.sizemin:
+            if entry.var_size() < args.sizemin:
                 continue
 
             key = entry.to_hash()

truvari/annotations/chunks.py

@@ -49,7 +49,7 @@ def chunks_main(args):
     """
     args = parse_args(args)
     v = truvari.VariantFile(args.input)
-    m = truvari.Matcher(args=args, pctseq=0)
+    m = truvari.VariantParams(args=args, pctseq=0)
     if args.bed:
         v = v.bed_fetch(args.bed)
     c = truvari.chunker(m, ('base', v))

truvari/annotations/hompct.py

@@ -45,7 +45,7 @@ def get_pct(chrom, start, end):
         tot = 0
         homs = 0
         for entry in v.fetch(chrom, max(0, start - args.buffer), min(v.header.contigs[chrom].length, end + args.buffer)):
-            if entry.size() > args.maxgt:
+            if entry.var_size() > args.maxgt:
                 continue
             if truvari.get_gt(entry.samples[0]["GT"]).name == "HOM":
                 homs += 1
@@ -66,7 +66,7 @@ def get_pct(chrom, start, end):
     out = truvari.VariantFile(args.output, 'w', header=header)
     v2 = truvari.VariantFile(args.input)
     for entry in v2:
-        if entry.size() >= args.minanno:
+        if entry.var_size() >= args.minanno:
             entry.translate(header)
             anno = get_pct(entry.chrom, *entry.boundaries())
             if anno is not None:

truvari/annotations/numneigh.py

@@ -132,7 +132,7 @@ def run(self):
         """
         last_pos = None
         for entry in self.in_vcf:
-            size = entry.size()
+            size = entry.var_size()
             if not last_pos:
                 last_pos = [entry.chrom, entry.start]
 

truvari/annotations/remap.py

@@ -117,7 +117,7 @@ def annotate_entry(self, entry):
         """
         Annotates entries in the vcf and writes to new vcf
         """
-        if entry.size() >= self.min_length:
+        if entry.var_size() >= self.min_length:
             entry.translate(self.n_header)
             remap, hits = self.remap_entry(entry)
             entry.info["REMAP"] = remap

truvari/annotations/repmask.py

@@ -73,7 +73,7 @@ def extract_seqs(self):
         fh = truvari.VariantFile(self.in_vcf)
         for pos, entry in enumerate(fh):
             tot_cnt += 1
-            entry_size = entry.size()
+            entry_size = entry.var_size()
             if self.min_length <= entry_size <= self.max_length:
                 cnt += 1
                 cntbp += entry_size
@@ -126,7 +126,7 @@ def annotate_entry(self, entry, hits):
         """
         best_hit_pct = 0
         best_hit = None
-        entry_size = entry.size()
+        entry_size = entry.var_size()
         for hit in hits:
             size_aln = abs(hit["RM_qstart"] - hit["RM_qend"]) + 1
             pct = size_aln / entry_size  # The TR that covers the most of the sequence

truvari/annotations/svinfo.py

@@ -42,7 +42,7 @@ def add_svinfo(entry, min_size=0, n_header=None):
         del entry.info['SVTYPE']
     if "SVLEN" in entry.info:
         del entry.info['SVLEN']
-    sz = entry.size()
+    sz = entry.var_size()
     if sz < min_size:
         return
     if n_header:

truvari/annotations/trf.py

@@ -181,7 +181,7 @@ def annotate(self, entry, score_filter=True):
         Figure out the hit and return
         """
         svtype = entry.var_type()
-        sz = entry.size()
+        sz = entry.var_size()
         repeat = []
         if svtype == truvari.SV.DEL:
             repeat = self.del_annotate(entry, sz, score_filter)
@@ -396,7 +396,7 @@ def process_ref_region(region, args):
                 continue
 
             svtype = entry.var_type()
-            svlen = entry.size()
+            svlen = entry.var_size()
             if svlen < args.min_length or svtype not in [truvari.SV.DEL, truvari.SV.INS]:
                 out.write(str(edit_entry(entry, None, new_header)))
                 continue
@@ -460,7 +460,7 @@ def process_tr_region(region, args):
             if not (entry.start >= region["start"] and entry.stop < region["end"]):
                 continue
             svtype = entry.var_type()
-            svlen = entry.size()
+            svlen = entry.var_size()
             if svlen < args.min_length or svtype not in [truvari.SV.DEL, truvari.SV.INS]:
                 out.write(str(edit_entry(entry, None, new_header)))
                 continue