add patent code

src/pyBiodatafuse/viz/patent_data.py

@@ -0,0 +1,84 @@
+# coding: utf-8
+
+"""Python file for extracting patent data from PubChem.
+
+The module contains special functions that are server expensive and can only be performed for smaller datasets.
+"""
+
+from typing import Optional, Tuple
+
+import pandas as pd
+import requests
+from tqdm import tqdm
+
+from pyBiodatafuse.utils import get_identifier_of_interest
+
+
+def get_patent_data(data_input: pd.DataFrame) -> dict:
+    """Get patent data summary from PubChem.
+
+    :param data_input: A dataframe with the BridgeDb or Pubchem harmonized output
+    :returns: A dictionary with the PubChem Compound ID as key and the patent counts as value
+    The output is the following:
+    >> {CID: ["US: X", "EP: X", "WO: X", "Others: X"]}
+    """
+    # Get column of interest
+    data_df = get_identifier_of_interest(data_input, "PubChem Compound")
+
+    cid_pat_dict = {}
+
+    # Neeed some pre-processing:
+    # 1. Remove duplicates (WO-03078408-A1 against WO03078408A1, Not classic Patent offices such as AR, AU againts WO)
+    # 2. Adding Granted, non-granted patent counts
+
+    for cid in tqdm(data_df["target"]):
+        patent_counter_dict = {"US": set(), "EP": set(), "WO": set(), "Others": set()}
+
+        patent_dict = requests.get(
+            f"https://pubchem.ncbi.nlm.nih.gov/rest/pug/compound/cid/{cid}/xrefs/PatentID/JSON"
+        ).json()
+
+        if "Fault" in patent_dict.keys():  # No patents found
+            cid_pat_dict[cid] = ""
+        else:
+            patents = patent_dict["InformationList"]["Information"][0]["PatentID"]
+
+            for patent in patents:
+                p = patent.replace("-", "")
+                if p.startswith("US"):
+                    patent_counter_dict["US"].add(p)
+                elif p.startswith("EP"):
+                    patent_counter_dict["EP"].add(p)
+                elif p.startswith("WO"):
+                    patent_counter_dict["WO"].add(p)
+                else:
+                    patent_counter_dict["Others"].add(p)
+
+        patent_counter_dict = [
+            f'{k}: {len(v)}' for k, v in patent_counter_dict.items() if len(v) > 0
+        ]
+
+        cid_pat_dict[cid] = patent_counter_dict
+
+    return cid_pat_dict
+
+
+def _process_data_for_plot(data_dict: dict) -> pd.DataFrame:
+    """Process data to map to plotting template."""
+
+    data = []
+
+    for cidx, pat_counter in data_dict.items():
+        for pat in pat_counter:
+            p, c = pat.split(": ")
+            data.append(
+                {
+                    "cid": cidx,
+                    "label": p,
+                    "value": c,
+                }
+            )
+
+    data_df = pd.DataFrame(data).set_index("cid")
+
+    return data_df