Merge pull request #272 from DeepRank/271_aminoacid_properties_dbodor

feat: improve component features
DeepRank · Dec 9, 2022 · 5229e9d · 5229e9d
2 parents bc02f79 + 426ba74
commit 5229e9d
Show file tree

Hide file tree

Showing 15 changed files with 399 additions and 287 deletions.
diff --git a/deeprankcore/domain/aminoacid_summary.xlsx b/deeprankcore/domain/aminoacid_summary.xlsx
diff --git a/deeprankcore/domain/aminoacidlist.py b/deeprankcore/domain/aminoacidlist.py
diff --git a/deeprankcore/domain/nodestorage.py b/deeprankcore/domain/nodestorage.py
@@ -8,16 +8,20 @@
 
 ## residue core features
 RESTYPE = "res_type" # AminoAcid object; former FEATURENAME_AMINOACID
-RESCHARGE = "res_charge" # float(<0); former FEATURENAME_CHARGE (was not assigned) 
-RESSIZE = "res_size" # int; former FEATURENAME_SIZE
+RESCHARGE = "charge" # float(<0); former FEATURENAME_CHARGE (was not assigned) 
 POLARITY = "polarity" #  Polarity object; former FEATURENAME_POLARITY
+RESSIZE = "res_size" # int; former FEATURENAME_SIZE
+RESMASS = "res_mass"
+RESPI = "res_pI"
 HBDONORS = "hb_donors" # int; former FEATURENAME_HYDROGENBONDDONORS
 HBACCEPTORS = "hb_acceptors"# int; former FEATURENAME_HYDROGENBONDACCEPTORS
 
 ## variant residue features
 VARIANTRES = "variant_res" # AminoAcid object; former FEATURENAME_VARIANTAMINOACID
 DIFFCHARGE = "diff_charge" # float
 DIFFSIZE = "diff_size" # int; former FEATURENAME_SIZEDIFFERENCE
+DIFFMASS = "diff_mass"
+DIFFPI = "diff_pI"
 DIFFPOLARITY = "diff_polarity" # [type?]; former FEATURENAME_POLARITYDIFFERENCE
 DIFFHBDONORS = "diff_hb_donors" # int; former FEATURENAME_HYDROGENBONDDONORSDIFFERENCE
 DIFFHBACCEPTORS = "diff_hb_acceptors" # int; former FEATURENAME_HYDROGENBONDACCEPTORSDIFFERENCE

diff --git a/deeprankcore/features/components.py b/deeprankcore/features/components.py
@@ -31,27 +31,32 @@ def add_features( # pylint: disable=unused-argument
 
         node.features[Nfeat.RESTYPE] = residue.amino_acid.onehot
         node.features[Nfeat.RESCHARGE] = residue.amino_acid.charge
-        node.features[Nfeat.RESSIZE] = residue.amino_acid.size
         node.features[Nfeat.POLARITY] = residue.amino_acid.polarity.onehot
-        node.features[Nfeat.HBDONORS] = residue.amino_acid.count_hydrogen_bond_donors
-        node.features[Nfeat.HBACCEPTORS] = residue.amino_acid.count_hydrogen_bond_acceptors
+        node.features[Nfeat.RESSIZE] = residue.amino_acid.size
+        node.features[Nfeat.RESMASS] = residue.amino_acid.mass
+        node.features[Nfeat.RESPI] = residue.amino_acid.pI
+        node.features[Nfeat.HBDONORS] = residue.amino_acid.hydrogen_bond_donors
+        node.features[Nfeat.HBACCEPTORS] = residue.amino_acid.hydrogen_bond_acceptors
 
         if single_amino_acid_variant is not None:
-
             wildtype = single_amino_acid_variant.wildtype_amino_acid
             variant = single_amino_acid_variant.variant_amino_acid
 
             if residue == single_amino_acid_variant.residue:
                 node.features[Nfeat.VARIANTRES] = variant.onehot
                 node.features[Nfeat.DIFFCHARGE] = variant.charge - wildtype.charge
-                node.features[Nfeat.DIFFSIZE] = variant.size - wildtype.size
                 node.features[Nfeat.DIFFPOLARITY] = variant.polarity.onehot - wildtype.polarity.onehot
-                node.features[Nfeat.DIFFHBDONORS] = variant.count_hydrogen_bond_donors - wildtype.count_hydrogen_bond_donors
-                node.features[Nfeat.DIFFHBACCEPTORS] = variant.count_hydrogen_bond_acceptors - wildtype.count_hydrogen_bond_acceptors
+                node.features[Nfeat.DIFFSIZE] = variant.size - wildtype.size
+                node.features[Nfeat.DIFFMASS] = variant.mass - wildtype.mass
+                node.features[Nfeat.DIFFPI] = variant.pI - wildtype.pI
+                node.features[Nfeat.DIFFHBDONORS] = variant.hydrogen_bond_donors - wildtype.hydrogen_bond_donors
+                node.features[Nfeat.DIFFHBACCEPTORS] = variant.hydrogen_bond_acceptors - wildtype.hydrogen_bond_acceptors
             else:
                 node.features[Nfeat.VARIANTRES] = residue.amino_acid.onehot
                 node.features[Nfeat.DIFFCHARGE] = 0
-                node.features[Nfeat.DIFFSIZE] = 0
                 node.features[Nfeat.DIFFPOLARITY] = np.zeros(residue.amino_acid.polarity.onehot.shape)
+                node.features[Nfeat.DIFFSIZE] = 0
+                node.features[Nfeat.DIFFMASS] = 0
+                node.features[Nfeat.DIFFPI] = 0
                 node.features[Nfeat.DIFFHBDONORS] = 0
                 node.features[Nfeat.DIFFHBACCEPTORS] = 0
diff --git a/deeprankcore/molstruct/aminoacid.py b/deeprankcore/molstruct/aminoacid.py
@@ -5,7 +5,7 @@
 class Polarity(Enum):
     "a value to express a residue's polarity"
 
-    APOLAR = 0
+    NONPOLAR = 0
     POLAR = 1
     NEGATIVE_CHARGE = 2
     POSITIVE_CHARGE = 3
@@ -26,35 +26,45 @@ def __init__( # pylint: disable=too-many-arguments
         name: str,
         three_letter_code: str,
         one_letter_code: str,
-        charge: float,
+        charge: int,
         polarity: Polarity,
         size: int,
-        count_hydrogen_bond_donors: int,
-        count_hydrogen_bond_acceptors: int,
+        mass: float,
+        pI: float, 
+        hydrogen_bond_donors: int,
+        hydrogen_bond_acceptors: int,
         index: int,
     ):
         """
         Args:
-            name(str): unique name for the amino acid
-            three_letter_code(str): code of the amino acid, as in PDB
-            one_letter_code(str): letter of the amino acid, as in fasta
-            charge(float, optional): the charge property of the amino acid
-            polarity(deeprank polarity enum, optional): the polarity property of the amino acid
-            size(int, optional): the number of heavy atoms in the side chain
-            index(int, optional): the rank of the amino acid, used for computing one-hot encoding
+            name (str): full name of the amino acid
+            three_letter_code (str): three-letter code of the amino acid (as in PDB)
+            one_letter_code (str): one-letter of the amino acid (as in fasta)
+            charge (int): charge of the amino acid
+            polarity (deeprank polarity enum): the polarity of the amino acid
+            size (int): the number of non-hydrogen atoms in the side chain
+            mass (float): average residue mass (i.e. mass of amino acid - H20) in Daltons
+            pI (float): isolectric point; pH at which the molecule has no net electric charge
+            hydrogen_bond_donors (int): number of hydrogen bond donors
+            hydrogen_bond_acceptors (int): number of hydrogen bond acceptors
+            index (int): the rank of the amino acid, used for computing one-hot encoding
         """
 
+        # amino acid nomenclature
         self._name = name
         self._three_letter_code = three_letter_code
         self._one_letter_code = one_letter_code
 
-        # these settings apply to the side chain
-        self._size = size
+        # side chain properties
         self._charge = charge
         self._polarity = polarity
-        self._count_hydrogen_bond_donors = count_hydrogen_bond_donors
-        self._count_hydrogen_bond_acceptors = count_hydrogen_bond_acceptors
+        self._size = size
+        self._mass = mass
+        self._pI = pI
+        self._hydrogen_bond_donors = hydrogen_bond_donors
+        self._hydrogen_bond_acceptors = hydrogen_bond_acceptors
 
+        # one hot encoding
         self._index = index
 
     @property
@@ -70,37 +80,45 @@ def one_letter_code(self) -> str:
         return self._one_letter_code
 
     @property
-    def onehot(self) -> np.ndarray:
-        if self._index is None:
-            raise ValueError(
-                "amino acid {self._name} index is not set, thus no onehot can be computed"
-            )
+    def charge(self) -> int:
+        return self._charge
 
-        # assumed that there are only 20 different amino acids
-        a = np.zeros(20)
-        a[self._index] = 1.0
+    @property
+    def polarity(self) -> Polarity:
+        return self._polarity
 
-        return a
+    @property
+    def size(self) -> int:
+        return self._size
 
     @property
-    def count_hydrogen_bond_donors(self) -> int:
-        return self._count_hydrogen_bond_donors
+    def mass(self) -> float:
+        return self._mass
 
     @property
-    def count_hydrogen_bond_acceptors(self) -> int:
-        return self._count_hydrogen_bond_acceptors
+    def pI(self) -> float:
+        return self._pI
 
     @property
-    def charge(self) -> float:
-        return self._charge
+    def hydrogen_bond_donors(self) -> int:
+        return self._hydrogen_bond_donors
 
     @property
-    def polarity(self) -> Polarity:
-        return self._polarity
+    def hydrogen_bond_acceptors(self) -> int:
+        return self._hydrogen_bond_acceptors
 
     @property
-    def size(self) -> int:
-        return self._size
+    def onehot(self) -> np.ndarray:
+        if self._index is None:
+            raise ValueError(
+                "amino acid {self._name} index is not set, thus no onehot can be computed"
+            )
+        # 20 canonical amino acids
+        # selenocysteine and pyrrolysine are indexed as cysteine and lysine, respectively
+        a = np.zeros(20)
+        a[self._index] = 1.0
+
+        return a
 
     @property
     def index(self) -> int:

diff --git a/deeprankcore/neuralnets/__init__.py b/deeprankcore/neuralnets/__init__.py
diff --git a/tests/data/hdf5/1ATN_ppi.hdf5 b/tests/data/hdf5/1ATN_ppi.hdf5