Make alleles optional in command-line predictors (default None)

mhctools/__init__.py

@@ -88,7 +88,7 @@ def __getattr__(name):
     raise AttributeError(
         "module %r has no attribute %r" % (__name__, name))
 
-__version__ = "3.31.6"
+__version__ = "3.31.7"
 
 __all__ = [
     "Prediction",

mhctools/base_commandline_predictor.py

@@ -834,6 +834,18 @@ def predict_pairs_dataframe(
             return pd.DataFrame(columns=COLUMNS)
         return pd.concat(dfs, ignore_index=True)
 
+    def _require_alleles(self):
+        """Guard the allele-consuming predict paths. alleles is optional at
+        construction so a predictor can score explicit (peptide, allele) pairs
+        via predict_pairs(); but predict()/predict_peptides() score against
+        self.alleles, and with none configured they would silently return
+        nothing (see _allele_groups). Fail loud instead."""
+        if not self.alleles:
+            raise ValueError(
+                "%s was constructed without alleles; pass alleles=[...] to the "
+                "constructor, or use predict_pairs(peptides, alleles) to score "
+                "explicit (peptide, allele) pairs." % type(self).__name__)
+
     def predict(self, peptides, n_flanks=None, c_flanks=None):
         """
         Predict for a list of peptide sequences.
@@ -842,6 +854,7 @@ def predict(self, peptides, n_flanks=None, c_flanks=None):
         available, parses directly to Pred objects. Otherwise falls back
         to converting from BindingPrediction.
         """
+        self._require_alleles()
         peptides, _, _ = self._check_flank_inputs(
             peptides, n_flanks, c_flanks)
         return self._predict_for_alleles(
@@ -850,6 +863,7 @@ def predict(self, peptides, n_flanks=None, c_flanks=None):
             allele_cli_names=getattr(self, "_allele_cli_names", None))
 
     def predict_peptides(self, peptides):
+        self._require_alleles()
         return self._predict_binding_predictions_for_alleles(
             peptides=peptides,
             alleles=self.alleles,

mhctools/base_predictor.py

@@ -112,7 +112,7 @@ class BasePredictor(object):
 
     def __init__(
             self,
-            alleles,
+            alleles=None,
             valid_alleles=None,
             default_peptide_lengths=None,
             min_peptide_length=8,
@@ -154,6 +154,11 @@ def __init__(
             against the tool's own supported-allele list, so that non-human
             alleles like H-2-Qa1 or BoLA-amani.1 can still be requested).
         """
+        # alleles is optional: a predictor built without one can still score
+        # explicit (peptide, allele) pairs via predict_pairs(). The
+        # allele-consuming predict()/predict_peptides() path guards on empty.
+        if alleles is None:
+            alleles = []
         # I find myself often constructing a predictor with just one allele
         # so as a convenience, allow user to not wrap that allele as a list
         if type(alleles) is str:

mhctools/netmhc.py

@@ -16,7 +16,7 @@
 from .netmhc3 import NetMHC3
 from .netmhc4 import NetMHC4
 
-def NetMHC(alleles,
+def NetMHC(alleles=None,
            default_peptide_lengths=[9],
            program_name="netMHC"):
     """

mhctools/netmhc3.py

@@ -16,7 +16,7 @@
 class NetMHC3(BaseCommandlinePredictor):
     def __init__(
             self,
-            alleles,
+            alleles=None,
             program_name="netMHC",
             default_peptide_lengths=[9]):
         BaseCommandlinePredictor.__init__(

mhctools/netmhc4.py

@@ -16,7 +16,7 @@
 class NetMHC4(BaseCommandlinePredictor):
     def __init__(
             self,
-            alleles,
+            alleles=None,
             program_name="netMHC",
             process_limit=0,
             default_peptide_lengths=[9]):

mhctools/netmhc_cons.py

@@ -16,7 +16,7 @@
 class NetMHCcons(BaseCommandlinePredictor):
     def __init__(
             self,
-            alleles,
+            alleles=None,
             program_name="netMHCcons",
             process_limit=0,
             default_peptide_lengths=[9]):

mhctools/netmhc_pan.py

@@ -47,7 +47,7 @@ def _parse_version(version_str):
 
 
 def NetMHCpan(
-        alleles,
+        alleles=None,
         program_name="netMHCpan",
         process_limit=-1,
         default_peptide_lengths=[9],

mhctools/netmhc_pan28.py

@@ -16,7 +16,7 @@
 class NetMHCpan28(BaseCommandlinePredictor):
     def __init__(
             self,
-            alleles,
+            alleles=None,
             default_peptide_lengths=[9],
             program_name="netMHCpan",
             process_limit=-1,

mhctools/netmhc_pan3.py

@@ -17,7 +17,7 @@
 class NetMHCpan3(BaseCommandlinePredictor):
     def __init__(
             self,
-            alleles,
+            alleles=None,
             default_peptide_lengths=[9],
             program_name="netMHCpan",
             process_limit=-1,

mhctools/netmhc_pan4.py

@@ -20,7 +20,7 @@
 class NetMHCpan4(BaseCommandlinePredictor):
     def __init__(
             self,
-            alleles,
+            alleles=None,
             default_peptide_lengths=[9],
             program_name="netMHCpan",
             process_limit=-1,
@@ -78,7 +78,7 @@ class NetMHCpan4_EL(NetMHCpan4):
     """
     def __init__(
             self,
-            alleles,
+            alleles=None,
             default_peptide_lengths=[9],
             program_name="netMHCpan",
             process_limit=-1,
@@ -98,7 +98,7 @@ class NetMHCpan4_BA(NetMHCpan4):
     """
     def __init__(
             self,
-            alleles,
+            alleles=None,
             default_peptide_lengths=[9],
             program_name="netMHCpan",
             process_limit=-1,

mhctools/netmhc_pan41.py

@@ -20,7 +20,7 @@
 class NetMHCpan41(BaseCommandlinePredictor):
     def __init__(
             self,
-            alleles,
+            alleles=None,
             default_peptide_lengths=[9],
             program_name="netMHCpan",
             process_limit=-1,
@@ -80,7 +80,7 @@ class NetMHCpan41_EL(NetMHCpan41):
     """
     def __init__(
             self,
-            alleles,
+            alleles=None,
             default_peptide_lengths=[9],
             program_name="netMHCpan",
             process_limit=-1,
@@ -101,7 +101,7 @@ class NetMHCpan41_BA(NetMHCpan41):
     """
     def __init__(
             self,
-            alleles,
+            alleles=None,
             default_peptide_lengths=[9],
             program_name="netMHCpan",
             process_limit=-1,

mhctools/netmhc_pan42.py

@@ -20,7 +20,7 @@
 class NetMHCpan42(BaseCommandlinePredictor):
     def __init__(
             self,
-            alleles,
+            alleles=None,
             default_peptide_lengths=[9],
             program_name="netMHCpan",
             process_limit=-1,
@@ -79,7 +79,7 @@ class NetMHCpan42_EL(NetMHCpan42):
     """NetMHCpan 4.2 in elution score mode."""
     def __init__(
             self,
-            alleles,
+            alleles=None,
             default_peptide_lengths=[9],
             program_name="netMHCpan",
             process_limit=-1,
@@ -98,7 +98,7 @@ class NetMHCpan42_BA(NetMHCpan42):
     """NetMHCpan 4.2 in binding affinity mode."""
     def __init__(
             self,
-            alleles,
+            alleles=None,
             default_peptide_lengths=[9],
             program_name="netMHCpan",
             process_limit=-1,

mhctools/netmhcii_pan.py

@@ -105,7 +105,7 @@ class NetMHCIIpan3(NetMHCIIpanBase):
     """
     def __init__(
             self,
-            alleles,
+            alleles=None,
             default_peptide_lengths=[15, 16, 17, 18, 19, 20],
             program_name="netMHCIIpan",
             process_limit=-1,
@@ -126,7 +126,7 @@ class NetMHCIIpan4(NetMHCIIpanBase):
     """
     def __init__(
             self,
-            alleles,
+            alleles=None,
             default_peptide_lengths=[15, 16, 17, 18, 19, 20],
             program_name="netMHCIIpan",
             process_limit=-1,
@@ -161,7 +161,7 @@ class NetMHCIIpan4_EL(NetMHCIIpan4):
     """
     def __init__(
             self,
-            alleles,
+            alleles=None,
             default_peptide_lengths=[15, 16, 17, 18, 19, 20],
             program_name="netMHCIIpan",
             process_limit=-1,
@@ -182,7 +182,7 @@ class NetMHCIIpan4_BA(NetMHCIIpan4):
     """
     def __init__(
             self,
-            alleles,
+            alleles=None,
             default_peptide_lengths=[15, 16, 17, 18, 19, 20],
             program_name="netMHCIIpan",
             process_limit=-1,
@@ -198,7 +198,7 @@ def __init__(
 
 
 def NetMHCIIpan(
-        alleles,
+        alleles=None,
         process_limit=-1,
         program_name="netMHCIIpan",
         default_peptide_lengths=[15, 16, 17, 18, 19, 20],
@@ -236,7 +236,7 @@ class NetMHCIIpan43(NetMHCIIpanBase):
     """
     def __init__(
             self,
-            alleles,
+            alleles=None,
             default_peptide_lengths=[15, 16, 17, 18, 19, 20],
             program_name="netMHCIIpan",
             process_limit=-1,
@@ -270,7 +270,7 @@ class NetMHCIIpan43_EL(NetMHCIIpan43):
     """
     def __init__(
             self,
-            alleles,
+            alleles=None,
             default_peptide_lengths=[15, 16, 17, 18, 19, 20],
             program_name="netMHCIIpan",
             process_limit=-1,
@@ -291,7 +291,7 @@ class NetMHCIIpan43_BA(NetMHCIIpan43):
     """
     def __init__(
             self,
-            alleles,
+            alleles=None,
             default_peptide_lengths=[15, 16, 17, 18, 19, 20],
             program_name="netMHCIIpan",
             process_limit=-1,

mhctools/netmhcstabpan.py

@@ -17,7 +17,7 @@
 class NetMHCstabpan(BaseCommandlinePredictor):
     def __init__(
             self,
-            alleles,
+            alleles=None,
             default_peptide_lengths=[],
             program_name="netMHCstabpan",
             process_limit=-1,

tests/test_commandline_predict_pairs.py

@@ -226,3 +226,36 @@ def test_predict_pairs_dataframe_skip_unsupported_all_unsupported_is_empty():
 
     assert list(df.columns) == list(COLUMNS)
     assert df.empty
+
+
+# ---- alleles optional at construction ---------------------------------------
+
+def test_base_predictor_alleles_optional_normalizes_to_empty():
+    from mhctools.base_predictor import BasePredictor
+
+    assert BasePredictor(default_peptide_lengths=[9]).alleles == []
+    assert BasePredictor(alleles=None, default_peptide_lengths=[9]).alleles == []
+    # an allele passed as a bare string still works
+    assert BasePredictor(
+        alleles="HLA-A*02:01", default_peptide_lengths=[9]).alleles == ["HLA-A*02:01"]
+
+
+def test_predict_pairs_works_without_constructor_alleles():
+    predictor = _PairStubPredictor()
+    predictor.alleles = []          # built for the pairs workflow, no default alleles
+
+    results = predictor.predict_pairs(
+        ["SIINFEKLL", "GILGFVFTL"], ["HLA-A*02:01", "HLA-B*07:02"])
+
+    assert [r.preds[0].peptide for r in results] == ["SIINFEKLL", "GILGFVFTL"]
+
+
+def test_predict_without_alleles_raises_rather_than_returning_nothing():
+    predictor = _PairStubPredictor()
+    predictor.alleles = []
+
+    with pytest.raises(ValueError, match="without alleles"):
+        predictor.predict(["SIINFEKLL"])
+    with pytest.raises(ValueError, match="without alleles"):
+        predictor.predict_peptides(["SIINFEKLL"])
+    assert predictor.calls == []    # guarded before running any command