From 4fe9a42a56bbcc3159bc01885963f13c0d03ec06 Mon Sep 17 00:00:00 2001
From: Darren Wight <darren@serenomica.com>
Date: Tue, 26 Aug 2025 12:41:50 +0200
Subject: [PATCH] feat: code to validate the NAT2 json made by scripts in teh
 pgx-engine added

---
 notebooks/validate_json.py | 142 +++++++++++++++++++++++++++++++++++++
 1 file changed, 142 insertions(+)
 create mode 100644 notebooks/validate_json.py

diff --git a/notebooks/validate_json.py b/notebooks/validate_json.py
new file mode 100644
index 0000000..7a1bef1
--- /dev/null
+++ b/notebooks/validate_json.py
@@ -0,0 +1,142 @@
+import marimo
+
+__generated_with = "0.14.16"
+app = marimo.App(width="medium")
+
+
+@app.cell
+def _():
+    import json
+    import marimo as mo
+    import pandas as pd
+    import numpy as np
+    return json, mo, np, pd
+
+
+@app.cell
+def _(mo):
+    mo.md(
+        r"""
+    # Validate the Json created by the pgx-pipeline
+
+    This notebook is to validate the json created by the scripts in the pgx-pipeline repo with the data from [pharmvar](https://www.pharmvar.org/).
+    """
+    )
+    return
+
+
+@app.cell
+def _(pd):
+    # original file downloaded from pharmvar for NAT2
+    pharmvar = pd.read_csv(
+        "../data/NAT2-6.2.15/GRCh38/NAT2.NC_000008.11.haplotypes.tsv",
+        sep="\t",
+        comment="#",
+    )
+    pharmvar.head(10)
+    return (pharmvar,)
+
+
+@app.cell
+def _(json):
+    # json created with the scripts in the PGx-engine using above file
+    with open("../data/NAT2_translation.json", "r") as handle:
+        json_data = json.load(handle)
+    print(json_data.keys())
+
+    # parse vars into list to use later for mapping
+    j_var_list = [
+        [str(j["position"]), j["rsid"], j["ref"], j["alts"]]
+        for j in json_data["variants"]
+    ]
+    print(j_var_list[0])
+    print(len(j_var_list))
+    return j_var_list, json_data
+
+
+@app.cell
+def _(j_var_list, json_data, np):
+    # extract the allele info for each allele and map with the variant details
+    allele_dict = {}
+    for allele in json_data["namedAlleles"]:
+        allele_id = allele["id"]
+
+        # if allele has no variants (i.e. wild type) give it a fixed entry
+        allele_dict[allele_id] = (
+            [
+                j_var_list[idx]
+                for idx, var in enumerate(allele["alleles"])
+                if var != None
+            ]
+            if allele_id not in ("NAT2*1", "NAT2*1.001")
+            else [[".", np.nan, np.nan, [np.nan]]]
+        )
+    allele_dict["NAT2*4.002"]
+    return (allele_dict,)
+
+
+@app.cell
+def _(allele_dict, pharmvar):
+    # check allele names match between json and the original file
+    pharmvar_hap_names = pharmvar["Haplotype Name"].unique().tolist()
+    json_hap_names = list(allele_dict.keys())
+
+    # check allele ids are same between json and pharvar sources
+    assert len(pharmvar_hap_names) == len(json_hap_names)
+    assert set(pharmvar_hap_names) == set(json_hap_names)
+    return (pharmvar_hap_names,)
+
+
+@app.cell
+def _(pharmvar, pharmvar_hap_names):
+    # parse the pharmvar table entries for each allele into a list for comparison with the json data
+    def get_pharmvar_vars(pharmvar_df, star_id):
+        vals = pharmvar_df.loc[pharmvar_df["Haplotype Name"] == star_id][
+            ["Variant Start", "rsID", "Reference Allele", "Variant Allele"]
+        ].values
+        return [[v[0], v[1], v[2], [v[3]]] for v in vals]
+
+
+    pharmvar_dict = {}
+    for name in pharmvar_hap_names:
+        pharmvar_dict[name] = get_pharmvar_vars(pharmvar, name)
+    pharmvar_dict["NAT2*4.002"]
+    return (pharmvar_dict,)
+
+
+@app.cell
+def _(allele_dict, pd, pharmvar_dict):
+    # compare the json to what is in the pharmvar
+    results = []
+    for pharm_name, pharm_data in pharmvar_dict.items():
+        if sorted(pharm_data) == sorted(allele_dict[pharm_name]):
+            # print(f"{pharm_name}\t\tMATCH")
+            results.append([pharm_name, "MATCH", ".", "."])
+        else:
+            # print([pharm_name, "NO MATCH", pharm_data, allele_dict[pharm_name]])
+            results.append(
+                [pharm_name, "NO MATCH", pharm_data, allele_dict[pharm_name]]
+            )
+    df = pd.DataFrame(
+        results, columns=["Haplotype Name", "Match", "PhamVar Data", "Json Data"]
+    )
+    df.value_counts("Match")
+    return
+
+
+@app.cell
+def _(mo):
+    mo.md(
+        r"""
+    ## Conclusions
+
+    - Alleles are all in the json from the Phamvar table.
+    - Allele definitions were correctly copied over.
+    - Scripts appear to make a correct form of the pharmvar data into json.
+    """
+    )
+    return
+
+
+if __name__ == "__main__":
+    app.run()