aiondata - pypi Package Compare versions

Comparing version 0.4.5 to 0.4.6

aiondata/bindingdb.py

@@ -0,1 +1,2 @@
+from collections import OrderedDict
 import io
@@ -10,2 +11,3 @@ from typing import Optional, Generator, Union, Tuple
 from .datasets import GeneratedDataset
+import polars as pl
 
@@ -21,12 +23,65 @@
     COLLECTION = "bindingdb"
+    SCHEMA = [
+        # Primary Identifiers
+        ("BindingDB Reactant_set_id", pl.Float64),
+        ("BindingDB MonomerID", pl.Float64),
+        ("BindingDB Entry DOI", pl.Utf8),
+        # Target-related Fields
+        ("Target Name", pl.Utf8),
+        ("UniProt (SwissProt) Primary ID of Target Chain", pl.Utf8),
+        ("UniProt (SwissProt) Secondary ID(s) of Target Chain", pl.Utf8),
+        ("UniProt (SwissProt) Alternative ID(s) of Target Chain", pl.Utf8),
+        ("UniProt (SwissProt) Recommended Name of Target Chain", pl.Utf8),
+        ("UniProt (SwissProt) Entry Name of Target Chain", pl.Utf8),
+        ("UniProt (TrEMBL) Primary ID of Target Chain", pl.Utf8),
+        ("UniProt (TrEMBL) Secondary ID(s) of Target Chain", pl.Utf8),
+        ("UniProt (TrEMBL) Alternative ID(s) of Target Chain", pl.Utf8),
+        ("UniProt (TrEMBL) Submitted Name of Target Chain", pl.Utf8),
+        ("UniProt (TrEMBL) Entry Name of Target Chain", pl.Utf8),
+        ("Target Source Organism According to Curator or DataSource", pl.Utf8),
+        ("BindingDB Target Chain Sequence", pl.Utf8),
+        # Ligand-related Fields
+        ("Ligand InChI", pl.Utf8),
+        ("Ligand InChI Key", pl.Utf8),
+        ("BindingDB Ligand Name", pl.Utf8),
+        ("SMILES", pl.Utf8),
+        ("PubChem CID of Ligand", pl.Float64),
+        ("PubChem SID of Ligand", pl.Float64),
+        ("ChEBI ID of Ligand", pl.Utf8),
+        ("ChEMBL ID of Ligand", pl.Utf8),
+        ("DrugBank ID of Ligand", pl.Utf8),
+        ("IUPHAR_GRAC ID of Ligand", pl.Utf8),
+        ("KEGG ID of Ligand", pl.Utf8),
+        ("ZINC ID of Ligand", pl.Utf8),
+        ("Ligand HET ID in PDB", pl.Utf8),
+        ("PDB ID(s) for Ligand-Target Complex", pl.Utf8),
+        ("PDB ID(s) of Target Chain", pl.Utf8),
+        # Interaction and Binding Affinity Measurements
+        ("Ki (nM)", pl.Float64),
+        ("IC50 (nM)", pl.Float64),
+        ("Kd (nM)", pl.Float64),
+        ("EC50 (nM)", pl.Float64),
+        ("kon (M-1-s-1)", pl.Float64),
+        ("koff (s-1)", pl.Float64),
+        # Experimental Conditions
+        ("pH", pl.Float64),
+        ("Temp C", pl.Float64),
+        # Miscellaneous
+        (
+            "Number of Protein Chains in Target (bigger than 1 implies a multichain complex)",
+            pl.Float64,
+        ),
+        # Links and References
+        ("Link to Ligand in BindingDB", pl.Utf8),
+        ("Link to Ligand-Target Pair in BindingDB", pl.Utf8),
+        ("From", pl.Utf8),
+        ("Curation/DataSource", pl.Utf8),
+        ("Article DOI", pl.Utf8),
+        ("PMID", pl.Utf8),
+        ("PubChem AID", pl.Utf8),
+        ("Patent Number", pl.Utf8),
+        ("Authors", pl.Utf8),
+        ("Institution", pl.Utf8),
+    ]
 
-    float_fields = {
-        "Ki (nM)",
-        "IC50 (nM)",
-        "Kd (nM)",
-        "EC50 (nM)",
-        "kon (M-1-s-1)",
-        "koff (s-1)",
-    }
-
     def __init__(self, fd: Optional[io.BufferedReader] = None):
@@ -62,3 +117,18 @@ """
         """
-        if prop_name in BindingDB.float_fields:
+
+        float_fields = {
+            name
+            for name, dtype in self.SCHEMA
+            if isinstance(dtype, (pl.Float64, pl.Float32))
+        }
+        if value == "":
+            return None
+        # Fudge numbers that are greater or less than a value
+        if value[0] == ">":
+            return float(value[1:]) * 1.01
+        if value[0] == "<":
+            return float(value[1:]) * 0.99
+        if "NV" in value:
+            return None
+        if prop_name in float_fields:
             try:
@@ -65,0 +135,0 @@ return float(value)

aiondata/datasets.py

@@ -68,2 +68,7 @@ import os
     def get_df(self) -> pl.DataFrame:
-        return pl.DataFrame(self.to_generator())
+        if self.SCHEMA is None:
+            return pl.DataFrame(
+                self.to_generator(), infer_schema_length=25000, strict=False
+            )
+        else:
+            return pl.DataFrame(self.to_generator(), schema=self.SCHEMA, strict=False)

PKG-INFO

 Metadata-Version: 2.1
 Name: aiondata
-Version: 0.4.5
+Version: 0.4.6
 Summary: A common data access layer for AI-driven drug discovery.
@@ -5,0 +5,0 @@ Home-page: https://www.github.com/aion-labs/aiondata

pyproject.toml

 [tool.poetry]
 name = "aiondata"
-version = "0.4.5"
+version = "0.4.6"
 description = "A common data access layer for AI-driven drug discovery."
@@ -5,0 +5,0 @@ authors = ["JJ Ben-Joseph <[e-mail]>"]

Alert delta unavailable

Currently unable to show alert delta for PyPI packages.

Improved metrics

Total package byte prevSize

53884

6.44%

Lines of code

655

12.93%