Merge c556009 into 042d1e4

mhcgnomes/__init__.py

@@ -16,7 +16,7 @@
 from .class2_pair import Class2Pair
 from .dataframe import dataframe_from_string_list, dataframe_from_parsed_objects
 from .errors import ParseError
-from .function_api import normalized_string, compact_string, parse
+from .function_api import cached_parser, parse
 from .gene import Gene
 from .haplotype import Haplotype
 from .mutation import Mutation
@@ -43,10 +43,8 @@
     "ParseError",
     "Serotype",
     "Species",
-
-    "compact_string",
     "dataframe_from_parsed_objects",
     "dataframe_from_string_list",
-    "normalized_string",
+    "cached_parser",
     "parse",
 ]

mhcgnomes/allele.py

@@ -83,7 +83,7 @@ def restrict_num_allele_fields(
         if num_fields >= self.num_allele_fields:
             return self
         else:
-            self.copy(
+            return self.copy(
                 allele_fields=self.allele_fields[:num_fields],
                 annotations=[] if drop_annotations else self.annotations,
                 mutations=[] if drop_mutations else self.mutations)

mhcgnomes/class2_pair.py

@@ -49,6 +49,10 @@ def get(cls, alpha, beta, raw_string=None):
             return None
         if beta is None:
             return None
+        if not isinstance(alpha, ResultWithMhcClass):
+            return None
+        if not isinstance(beta, ResultWithMhcClass):
+            return None
         if alpha.mhc_class == beta.mhc_class:
             mhc_class = alpha.mhc_class
         else:

mhcgnomes/data/haplotypes.yaml

@@ -39,7 +39,7 @@ H2:
     - Aa*b
     - Ab*b
     - Eb*b
-    - IE*k
+    # Ea: null
     - Qa2*b
     - Qa1*a
   bc:

mhcgnomes/function_api.py

@@ -44,7 +44,8 @@ def parse(
         map_allele_aliases=MAP_ALLELE_ALIASES,
         infer_class2_pairing=INFER_CLASS2_PAIRING,
         simplify_haplotypes_if_possible=COLLAPSE_SINGLETON_HAPLOTYPES,
-        valid_result_types=[],
+        required_result_types=[],
+        preferred_result_types=[],
         raise_on_error=True):
     """
     Parse MHC alleles into a structured representation.
@@ -68,9 +69,12 @@ def parse(
         If a Haplotype contains only a single allele or Class II allele pair,
         then return the allele instead of a haplotype.
 
-    valid_result_types : list of class
+    required_result_types : list of type
         Only return results of the given classes.
 
+    preferred_result_types : list of type
+        Return a result that's one of these classes if possible, otherwise None.
+
     raise_on_error : bool
         Raise an exception if string can't be parsed. If False, return None
         instead.
@@ -84,129 +88,5 @@ def parse(
         default_species=default_species,
         infer_class2_pairing=infer_class2_pairing,
         raise_on_error=raise_on_error,
-        valid_result_types=valid_result_types)
-
-def normalized_string(
-        raw_string,
-        include_species_prefix=True,
-        use_old_species_prefix=False,
-        map_allele_aliases=True,
-        simplify_haplotypes_if_possible=COLLAPSE_SINGLETON_HAPLOTYPES,
-        infer_class2_pairing=True,
-        default_species=DEFAULT_SPECIES_PREFIX,
-        valid_result_types=[],
-        raise_on_error=True):
-    """
-    Transform MHC alleles into a canonical string representation.
-
-    Examples:
-        A2 -> HLA-A2
-        A0201 -> HLA-A*02:01
-        H2-K-k -> H2-Kk
-        RT-1*9.5:f -> RT1-9.5f
-        DRB1_0101 -> HLA-DRB1*01:01
-
-    Parameters
-    ----------
-    raw_string : str
-        String corresponding to allele, locus, or other MHC-related name
-
-    include_species_prefix : bool
-        Include species in the normalized. If False, then you would
-        get "A*02:01" for "A0201", instead of "HLA-A*02:01"
-
-    use_old_species_prefix : bool
-        For species which have a newer four-digit code and and older locus
-        name (such as "Ecqa" / "ELA"), use the older species prefix in the
-        result.
-
-    map_allele_aliases : bool
-
-    simplify_haplotypes_if_possible : bool
-        If a Haplotype contains only a single allele or Class II allele pair,
-        then return the allele instead of a haplotype.
-
-    infer_class2_pairing : bool
-        If given only the alpha or beta chain of a Class II allele,
-        try to infer the most likely pairing from population frequencies.
-
-    default_species_prefix : str
-        By default, parse alleles like "A*02:01" as human but it's possible
-        to change this to some other species.
-
-    valid_result_types : list of class
-        Only return results of the given classes.
-
-    raise_on_error : bool
-        Raise an exception if string can't be parsed. If False, return None
-        instead.
-    """
-    parsed_object = parse(
-        raw_string,
-        infer_class2_pairing=infer_class2_pairing,
-        default_species=default_species,
-        map_allele_aliases=map_allele_aliases,
-        raise_on_error=raise_on_error,
-        simplify_haplotypes_if_possible=simplify_haplotypes_if_possible,
-        valid_result_types=valid_result_types)
-    if not parsed_object:
-        return None
-    return parsed_object.to_string(
-        include_species=include_species_prefix,
-        use_old_species_prefix=use_old_species_prefix)
-
-def compact_string(
-        raw_string,
-        use_old_species_prefix=False,
-        map_allele_aliases=True,
-        simplify_haplotypes_if_possible=COLLAPSE_SINGLETON_HAPLOTYPES,
-        infer_class2_pairing=False,
-        default_species="HLA",
-        valid_result_types=[],
-        raise_on_error=True):
-    """
-    Turn HLA-A*02:01 into A0201 or H-2-D-b into H-2Db or
-    HLA-DPA1*01:05-DPB1*100:01 into DPA10105-DPB110001
-
-    Parameters
-    ----------
-    raw_string : str
-        String corresponding to allele, locus, or other MHC-related name
-
-    use_old_species_prefix : bool
-        For species which have a newer four-digit code and and older locus
-        name (such as "Ecqa" / "ELA"), use the older species prefix in the
-        result.
-
-    map_allele_aliases : bool
-
-    simplify_haplotypes_if_possible : bool
-        If a Haplotype contains only a single allele or Class II allele pair,
-        then return the allele instead of a haplotype.
-
-    infer_class2_pairing : bool
-        If given only the alpha or beta chain of a Class II allele,
-        try to infer the most likely pairing from population frequencies.
-
-    default_species : str
-        By default, parse alleles like "A*02:01" as human but it's possible
-        to change this to some other species.
-
-    valid_result_types : list of class
-        Only return results of the given classes.
-
-    raise_on_error : bool
-        Raise an exception if string can't be parsed. If False, return None
-        instead.
-    """
-    parsed_object = parse(
-        raw_string,
-        map_allele_aliases=map_allele_aliases,
-        infer_class2_pairing=infer_class2_pairing,
-        default_species=default_species,
-        raise_on_error=raise_on_error,
-        simplify_haplotypes_if_possible=simplify_haplotypes_if_possible,
-        valid_result_types=valid_result_types)
-    if not parsed_object:
-        return None
-    return parsed_object.compact_string(use_old_species_prefix=use_old_species_prefix)
+        required_result_types=required_result_types,
+        preferred_result_types=preferred_result_types)

mhcgnomes/haplotype.py

@@ -17,12 +17,12 @@
     is_valid_restriction,
     restrict_alleles,
 )
-from .result_with_species import ResultWithSpecies
 from .class2_locus import Class2Locus
 from .class2_pair import Class2Pair
+from .result_with_multiple_alleles  import ResultWithMultipleAlleles
 from .species import Species
 
-class Haplotype(ResultWithSpecies):
+class Haplotype(ResultWithMultipleAlleles):
     def __init__(
             self,
             species : Species,
@@ -32,12 +32,12 @@ def __init__(
             locus_restriction : Union[Class2Locus, None] = None,
             parent_haplotypes : Union[Sequence["Haplotype"], None] = None,
             raw_string : Union[str, None] = None):
-        ResultWithSpecies.__init__(
+        ResultWithMultipleAlleles.__init__(
             self,
             species=species,
+            name=name,
+            alleles=alleles,
             raw_string=raw_string)
-        self.name = name
-        self.alleles = tuple(sorted(alleles))
         self.class_restriction = class_restriction
         self.locus_restriction = locus_restriction
         self.parent_haplotypes = parent_haplotypes
@@ -75,10 +75,6 @@ def eq_field_names(cls):
     def haplotype_name(self):
         return self.name
 
-    @property
-    def num_alleles(self):
-        return len(self.alleles)
-
     def restrict_mhc_class(self, class_restriction, raise_on_error=True):
         if class_restriction is None:
             return self
@@ -192,23 +188,3 @@ def to_string(
 
     def compact_string(self, include_species=False):
         return self.to_string(include_species=include_species)
-
-    @property
-    def has_allele(self):
-        return len(self.alleles) > 0
-
-    @property
-    def is_class1(self):
-        return all([allele.is_class1 for allele in self.alleles])
-
-    @property
-    def is_class2(self):
-        return all([allele.is_class2 for allele in self.alleles])
-
-    @property
-    def is_class2_alpha(self):
-        return all([allele.is_class2_alpha for allele in self.alleles])
-
-    @property
-    def is_class2_beta(self):
-        return all([allele.is_class2_beta for allele in self.alleles])

mhcgnomes/mhc_class_helpers.py

@@ -77,12 +77,19 @@ def restrict_alleles(alleles, mhc_class):
     ]
 
 
-def normalize_mhc_class_string(mhc_class):
+def normalize_mhc_class_string(mhc_class, raise_on_error=True):
     original_string = mhc_class
     mhc_class = mhc_class.lower()
+    if mhc_class.startswith("class-"):
+        _, mhc_class = mhc_class.split("class-")
+    elif mhc_class.startswith("class "):
+        _, mhc_class = mhc_class.split("class ")
     mhc_class = mhc_class.replace("i", "I")
     mhc_class = mhc_class.replace("1", "I")
     mhc_class = mhc_class.replace("2", "II")
     if mhc_class not in valid_class_restrictions:
-        raise ParseError("Invalid MHC class: '%s'" % original_string)
+        if raise_on_error:
+            raise ParseError("Invalid MHC class: '%s'" % original_string)
+        else:
+            return None
     return mhc_class