Upload 18 files

README.md

@@ -1,5 +1,67 @@
----
-license: other
-license_name: cc0
-license_link: https://creativecommons.org/publicdomain/zero/1.0/
----
+# QM9 Molecular Data Preprocessing and ML Pipeline
+
+This repository contains a series of scripts for processing the QM9 dataset, computing molecular descriptors (e.g., ECFP4), cleaning and organizing molecular data, and preparing it for machine learning (ML) tasks.
+
+
+This pipeline transforms raw .xyz molecular files into a structured ML-ready format by:
+	1.	Extracting properties
+	2.	Calculating ECFP4 fingerprints
+	3.	Cleaning and filtering problematic entries
+	4.	Organizing columns and unpacking features
+	5.	Finalizing the dataset for ML
+	6.	Splitting it into training, validation, and test sets
+
+# Scripts
+
+| File                     | Description |
+|--------------------------|-------------|
+| `00_download_data.sh`    | Downloads, unzips, and moves the files into a directory called `input_data`. |
+| `01_batch_data.py`       | Creates many batch files, each containing a portion of the data as file paths. Batching helps parallelize the next step. |
+| `02_process_batch.py`    | Reads and parses `.xyz` files, saving processed molecular data (e.g., atoms, coordinates, properties) in `.parquet` format. |
+| `03_run_batches.sh`      | A shell script to process batches of `.xyz` files using SLURM or local runs. Calls `02_process_batch.py` over all batch files. |
+| `04_merge_data.py`       | Merges all batch-level `.parquet` outputs into a single unified dataset. |
+| `05_sanitize_data.py`    | Standardizes SMILES strings and removes molecules that fail sanitization. |
+| `06_calculate_ecfp.py`   | Computes ECFP4 molecular fingerprints using RDKit. |
+| `07_calculate_properties.py` | Calculates and appends molecular properties to the dataset. |
+| `08_organize_columns.py` | Reorganizes column order, drops unnecessary columns, and prepares a clean structure. |
+| `09_unpack_ecfp4.py`     | Unpacks the ECFP4 bit vectors into individual binary columns for ML input. |
+| `10_finalize_ml_data.py` | Final checks and formatting to ensure the data is ready for ML training. |
+| `11_split_ml_data.py`    | Splits the final dataset into train/val/test sets using fixed proportions (e.g., 80/10/10). |
+
+
+
+
+
+
+---
+
+ # Output
+
+
+
+- Clean .parquet file containing ECFP4 + molecular properties, Split files for ML training: train.parquet, val.parquet, test.parquet
+
+### output format
+
+| Property Group (Column Range)                         | Columns                                                                                                                                                             |
+|--------------------------------------------------------|----------------------------------------------------------------------------------------------------------------------------------------------------------------------|
+| Identifier (1–1)                                       | `ID`                                                                                                                                                                 |
+| Chemical descriptors (2–9)                             | `MolWt`, `ClogP`, `TPSA`, `HBD`, `HBA`, `NumRotatableBonds`, `RingCount`, `FractionCSP3`                                                                             |
+| Quantum-chemical & thermodynamic properties (10–24)    | `Rotational_Constant_A`, `Rotational_Constant_B`, `Rotational_Constant_C`, `Dipole_Moment`, `Isotropic_polarizability`, `Energy_of_HOMO`, `Energy_of_LUMO`, `LUMO_HOMO_GAP`, `Electronic_spatial_extent`, `Zero_point_vibrational_energy`, `Internal_energy_at_0_K`, `Internal_energy_at_298.15_K`, `Enthalpy_at_298.15_K`, `Free_energy_at_298.15_K`, `Heat_capacity_at_298.15_K` |
+| Structural fingerprint (25–25)                         | `Ecfp_4` (2,048-bit Morgan fingerprint, radius 2)                                                                                                                    |
+
+
+
+
+# Sources: 
+
+    L. Ruddigkeit, R. van Deursen, L. C. Blum, J.-L. Reymond, Enumeration of 166 billion organic small molecules in the chemical universe database GDB-17, J. Chem. Inf. Model. 52, 2864–2875, 2012.
+    R. Ramakrishnan, P. O. Dral, M. Rupp, O. A. von Lilienfeld, Quantum chemistry structures and properties of 134 kilo molecules, Scientific Data 1, 140022, 2014. 
+
+
+# License
+
+This project is licensed under CC0 (for use with QM9 which is public domain). Attribution is appreciated but not required.
+
+
+

data/QML9.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:0a11c987215e55ba1aaaf918cd1bdc78c9f727ff3046fc7cfcf64880eddcfdad
+size 18136924

data/test_split.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:4d4c6426bd2f1bad91dfc8f6cfded2b16ee09af9bd37ec34c763de3201f53ecd
+size 3563047

data/train_split.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:b13ce66c1cfbf776e512a641e90cb39f010c1e7badcf0440fce291ff8c276d1e
+size 17905474

data/validation_split.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:6271827abe5802361841942e4c8f23dcb37427376261c5b3efb8fb8277f3c8a8
+size 3566196

src/00_download_data.sh

@@ -0,0 +1,5 @@
+#!/bin/bash
+wget https://figshare.com/ndownloader/files/3195389
+tar -xvjf 3195389
+mkdir input_data
+mv *.xyz input_data/

src/01_batch_data.py

@@ -0,0 +1,20 @@
+import os
+import math
+
+input_dir = "input_data"        
+output_dir = "batches"           
+batch_size = 1000                
+os.makedirs(output_dir, exist_ok=True)
+
+all_files = sorted([
+    os.path.join(input_dir, f)
+    for f in os.listdir(input_dir)
+    if f.endswith(".xyz")
+])
+
+n_batches = math.ceil(len(all_files) / batch_size)
+
+for i in range(n_batches):
+    batch_files = all_files[i*batch_size : (i+1)*batch_size]
+    with open(f"{output_dir}/batch_{i:03d}.txt", "w") as f:
+        f.write("\n".join(batch_files))

src/02_process_batch.py

@@ -0,0 +1,88 @@
+import pandas as pd
+import argparse
+import os
+
+def safe_float(value):
+    try:
+        # Handle malformed scientific notation like 2.1997*^-6
+        clean_value = value.replace("*^-", "e-").replace("*^", "e")
+        return float(clean_value)
+    except Exception as e:
+        raise ValueError(f"Failed to parse float from '{value}': {e}")
+
+def process_xyz(filepath):
+    data = {}
+    with open(filepath, 'r') as f:
+        lines = f.readlines()
+
+    data["n_atoms"] = int(lines[0].strip())
+    values = lines[1].split()
+    data["ID"] = values[1]
+
+    #data["SMILES_GDB17"], data["SMILES_B3LYP"] = lines[-2].strip().split()
+    data["SMILES_GDB17"] = lines[-2].strip().split()[0]
+
+
+    data["Rotational_Constant_A"] = safe_float(values[2])
+    data["Rotational_Constant_B"] = safe_float(values[3])
+    data["Rotational_Constant_C"] = safe_float(values[4])
+    data["Dipole_Moment"] = safe_float(values[5])
+    data["Isotropic_polarizability"] = safe_float(values[6])
+    data["Energy_of_HOMO"] = safe_float(values[7])
+    data["Energy_of_LUMO"] = safe_float(values[8])
+    data["LUMO_HOMO_GAP"] = safe_float(values[9])
+    data["Electronic_spatial_extent"] = safe_float(values[10])
+    data["Zero_point_vibrational_energy"] = safe_float(values[11])
+    data["Internal_energy_at_0_K"] = safe_float(values[12])
+    data["Internal_energy_at_298.15_K"] = safe_float(values[13])
+    data["Enthalpy_at_298.15_K"] = safe_float(values[14])
+    data["Free_energy_at_298.15_K"] = safe_float(values[15])
+    data["Heat_capacity_at_298.15_K"] = safe_float(values[16])
+
+    for i in range(data["n_atoms"]):
+        atom = lines[2 + i].split()
+        data[f"element_{i}"] = atom[0]
+        data[f"x_{i}"] = safe_float(atom[1])
+        data[f"y_{i}"] = safe_float(atom[2])
+        data[f"z_{i}"] = safe_float(atom[3])
+        data[f"charge_{i}"] = safe_float(atom[4])
+
+
+    return data
+
+
+def main():
+    parser = argparse.ArgumentParser()
+    parser.add_argument("--file_list", required=True, help="Path to .txt file listing input .xyz files")
+    parser.add_argument("--output_file", required=True, help="Path to output .parquet file (checkpointed)")
+    parser.add_argument("--checkpoint_every", type=int, default=100, help="Save after every N molecules")
+    args = parser.parse_args()
+    print("Processing",args.file_list)
+    with open(args.file_list, 'r') as f:
+        files = [line.strip() for line in f if line.strip()]
+
+    records = {}
+    for i, path in enumerate(files):
+        try:
+            data = process_xyz(path)
+            n_atoms=data["n_atoms"]
+            records[n_atoms]=data
+        except Exception as e:
+            print(f" Failed on {path}: {e}")
+
+        if (i + 1) % args.checkpoint_every == 0:
+            df = pd.DataFrame(records)
+            df.to_parquet(args.output_file, index=False)
+            print(f"Checkpointed {len(df)} molecules at {i+1}/{len(files)}")
+
+    # Final save
+    if records:
+        df = pd.DataFrame(records)
+        df.to_parquet(args.output_file, index=False)
+        print(f" Final write: {len(df)} molecules → {args.output_file}")
+    else:
+        print(" No valid data to save.")
+
+
+if __name__ == "__main__":
+    main()

src/03_run_batches.sh

@@ -0,0 +1,91 @@
+#!/bin/bash
+python src/02_process_batch.py --file_list ./batches/batch_000.txt --output_file checkpoints/batch_000.parquet --checkpoint_every 100
+python src/02_process_batch.py --file_list ./batches/batch_001.txt --output_file checkpoints/batch_001.parquet --checkpoint_every 100
+python src/02_process_batch.py --file_list ./batches/batch_002.txt --output_file checkpoints/batch_002.parquet --checkpoint_every 100
+python src/02_process_batch.py --file_list ./batches/batch_003.txt --output_file checkpoints/batch_003.parquet --checkpoint_every 100
+python src/02_process_batch.py --file_list ./batches/batch_004.txt --output_file checkpoints/batch_004.parquet --checkpoint_every 100
+python src/02_process_batch.py --file_list ./batches/batch_005.txt --output_file checkpoints/batch_005.parquet --checkpoint_every 100
+python src/02_process_batch.py --file_list ./batches/batch_006.txt --output_file checkpoints/batch_006.parquet --checkpoint_every 100
+python src/02_process_batch.py --file_list ./batches/batch_007.txt --output_file checkpoints/batch_007.parquet --checkpoint_every 100
+python src/02_process_batch.py --file_list ./batches/batch_008.txt --output_file checkpoints/batch_008.parquet --checkpoint_every 100
+python src/02_process_batch.py --file_list ./batches/batch_009.txt --output_file checkpoints/batch_009.parquet --checkpoint_every 100
+python src/02_process_batch.py --file_list ./batches/batch_010.txt --output_file checkpoints/batch_010.parquet --checkpoint_every 100
+python src/02_process_batch.py --file_list ./batches/batch_011.txt --output_file checkpoints/batch_011.parquet --checkpoint_every 100
+python src/02_process_batch.py --file_list ./batches/batch_012.txt --output_file checkpoints/batch_012.parquet --checkpoint_every 100
+python src/02_process_batch.py --file_list ./batches/batch_013.txt --output_file checkpoints/batch_013.parquet --checkpoint_every 100
+python src/02_process_batch.py --file_list ./batches/batch_014.txt --output_file checkpoints/batch_014.parquet --checkpoint_every 100
+python src/02_process_batch.py --file_list ./batches/batch_015.txt --output_file checkpoints/batch_015.parquet --checkpoint_every 100
+python src/02_process_batch.py --file_list ./batches/batch_016.txt --output_file checkpoints/batch_016.parquet --checkpoint_every 100
+python src/02_process_batch.py --file_list ./batches/batch_017.txt --output_file checkpoints/batch_017.parquet --checkpoint_every 100
+python src/02_process_batch.py --file_list ./batches/batch_018.txt --output_file checkpoints/batch_018.parquet --checkpoint_every 100
+python src/02_process_batch.py --file_list ./batches/batch_019.txt --output_file checkpoints/batch_019.parquet --checkpoint_every 100
+python src/02_process_batch.py --file_list ./batches/batch_020.txt --output_file checkpoints/batch_020.parquet --checkpoint_every 100
+python src/02_process_batch.py --file_list ./batches/batch_021.txt --output_file checkpoints/batch_021.parquet --checkpoint_every 100
+python src/02_process_batch.py --file_list ./batches/batch_022.txt --output_file checkpoints/batch_022.parquet --checkpoint_every 100
+python src/02_process_batch.py --file_list ./batches/batch_023.txt --output_file checkpoints/batch_023.parquet --checkpoint_every 100
+python src/02_process_batch.py --file_list ./batches/batch_024.txt --output_file checkpoints/batch_024.parquet --checkpoint_every 100
+python src/02_process_batch.py --file_list ./batches/batch_025.txt --output_file checkpoints/batch_025.parquet --checkpoint_every 100
+python src/02_process_batch.py --file_list ./batches/batch_026.txt --output_file checkpoints/batch_026.parquet --checkpoint_every 100
+python src/02_process_batch.py --file_list ./batches/batch_027.txt --output_file checkpoints/batch_027.parquet --checkpoint_every 100
+python src/02_process_batch.py --file_list ./batches/batch_028.txt --output_file checkpoints/batch_028.parquet --checkpoint_every 100
+python src/02_process_batch.py --file_list ./batches/batch_029.txt --output_file checkpoints/batch_029.parquet --checkpoint_every 100
+python src/02_process_batch.py --file_list ./batches/batch_030.txt --output_file checkpoints/batch_030.parquet --checkpoint_every 100
+python src/02_process_batch.py --file_list ./batches/batch_031.txt --output_file checkpoints/batch_031.parquet --checkpoint_every 100
+python src/02_process_batch.py --file_list ./batches/batch_032.txt --output_file checkpoints/batch_032.parquet --checkpoint_every 100
+python src/02_process_batch.py --file_list ./batches/batch_033.txt --output_file checkpoints/batch_033.parquet --checkpoint_every 100
+python src/02_process_batch.py --file_list ./batches/batch_034.txt --output_file checkpoints/batch_034.parquet --checkpoint_every 100
+python src/02_process_batch.py --file_list ./batches/batch_035.txt --output_file checkpoints/batch_035.parquet --checkpoint_every 100
+python src/02_process_batch.py --file_list ./batches/batch_036.txt --output_file checkpoints/batch_036.parquet --checkpoint_every 100
+python src/02_process_batch.py --file_list ./batches/batch_037.txt --output_file checkpoints/batch_037.parquet --checkpoint_every 100
+python src/02_process_batch.py --file_list ./batches/batch_038.txt --output_file checkpoints/batch_038.parquet --checkpoint_every 100
+python src/02_process_batch.py --file_list ./batches/batch_039.txt --output_file checkpoints/batch_039.parquet --checkpoint_every 100
+python src/02_process_batch.py --file_list ./batches/batch_040.txt --output_file checkpoints/batch_040.parquet --checkpoint_every 100
+python src/02_process_batch.py --file_list ./batches/batch_041.txt --output_file checkpoints/batch_041.parquet --checkpoint_every 100
+python src/02_process_batch.py --file_list ./batches/batch_042.txt --output_file checkpoints/batch_042.parquet --checkpoint_every 100
+python src/02_process_batch.py --file_list ./batches/batch_043.txt --output_file checkpoints/batch_043.parquet --checkpoint_every 100
+python src/02_process_batch.py --file_list ./batches/batch_044.txt --output_file checkpoints/batch_044.parquet --checkpoint_every 100
+python src/02_process_batch.py --file_list ./batches/batch_045.txt --output_file checkpoints/batch_045.parquet --checkpoint_every 100
+python src/02_process_batch.py --file_list ./batches/batch_046.txt --output_file checkpoints/batch_046.parquet --checkpoint_every 100
+python src/02_process_batch.py --file_list ./batches/batch_047.txt --output_file checkpoints/batch_047.parquet --checkpoint_every 100
+python src/02_process_batch.py --file_list ./batches/batch_048.txt --output_file checkpoints/batch_048.parquet --checkpoint_every 100
+python src/02_process_batch.py --file_list ./batches/batch_049.txt --output_file checkpoints/batch_049.parquet --checkpoint_every 100
+python src/02_process_batch.py --file_list ./batches/batch_050.txt --output_file checkpoints/batch_050.parquet --checkpoint_every 100
+python src/02_process_batch.py --file_list ./batches/batch_051.txt --output_file checkpoints/batch_051.parquet --checkpoint_every 100
+python src/02_process_batch.py --file_list ./batches/batch_052.txt --output_file checkpoints/batch_052.parquet --checkpoint_every 100
+python src/02_process_batch.py --file_list ./batches/batch_053.txt --output_file checkpoints/batch_053.parquet --checkpoint_every 100
+python src/02_process_batch.py --file_list ./batches/batch_054.txt --output_file checkpoints/batch_054.parquet --checkpoint_every 100
+python src/02_process_batch.py --file_list ./batches/batch_055.txt --output_file checkpoints/batch_055.parquet --checkpoint_every 100
+python src/02_process_batch.py --file_list ./batches/batch_056.txt --output_file checkpoints/batch_056.parquet --checkpoint_every 100
+python src/02_process_batch.py --file_list ./batches/batch_057.txt --output_file checkpoints/batch_057.parquet --checkpoint_every 100
+python src/02_process_batch.py --file_list ./batches/batch_058.txt --output_file checkpoints/batch_058.parquet --checkpoint_every 100
+python src/02_process_batch.py --file_list ./batches/batch_059.txt --output_file checkpoints/batch_059.parquet --checkpoint_every 100
+python src/02_process_batch.py --file_list ./batches/batch_060.txt --output_file checkpoints/batch_060.parquet --checkpoint_every 100
+python src/02_process_batch.py --file_list ./batches/batch_061.txt --output_file checkpoints/batch_061.parquet --checkpoint_every 100
+python src/02_process_batch.py --file_list ./batches/batch_062.txt --output_file checkpoints/batch_062.parquet --checkpoint_every 100
+python src/02_process_batch.py --file_list ./batches/batch_063.txt --output_file checkpoints/batch_063.parquet --checkpoint_every 100
+python src/02_process_batch.py --file_list ./batches/batch_064.txt --output_file checkpoints/batch_064.parquet --checkpoint_every 100
+python src/02_process_batch.py --file_list ./batches/batch_065.txt --output_file checkpoints/batch_065.parquet --checkpoint_every 100
+python src/02_process_batch.py --file_list ./batches/batch_066.txt --output_file checkpoints/batch_066.parquet --checkpoint_every 100
+python src/02_process_batch.py --file_list ./batches/batch_067.txt --output_file checkpoints/batch_067.parquet --checkpoint_every 100
+python src/02_process_batch.py --file_list ./batches/batch_068.txt --output_file checkpoints/batch_068.parquet --checkpoint_every 100
+python src/02_process_batch.py --file_list ./batches/batch_069.txt --output_file checkpoints/batch_069.parquet --checkpoint_every 100
+python src/02_process_batch.py --file_list ./batches/batch_070.txt --output_file checkpoints/batch_070.parquet --checkpoint_every 100
+python src/02_process_batch.py --file_list ./batches/batch_071.txt --output_file checkpoints/batch_071.parquet --checkpoint_every 100
+python src/02_process_batch.py --file_list ./batches/batch_072.txt --output_file checkpoints/batch_072.parquet --checkpoint_every 100
+python src/02_process_batch.py --file_list ./batches/batch_073.txt --output_file checkpoints/batch_073.parquet --checkpoint_every 100
+python src/02_process_batch.py --file_list ./batches/batch_074.txt --output_file checkpoints/batch_074.parquet --checkpoint_every 100
+python src/02_process_batch.py --file_list ./batches/batch_075.txt --output_file checkpoints/batch_075.parquet --checkpoint_every 100
+python src/02_process_batch.py --file_list ./batches/batch_076.txt --output_file checkpoints/batch_076.parquet --checkpoint_every 100
+python src/02_process_batch.py --file_list ./batches/batch_077.txt --output_file checkpoints/batch_077.parquet --checkpoint_every 100
+python src/02_process_batch.py --file_list ./batches/batch_078.txt --output_file checkpoints/batch_078.parquet --checkpoint_every 100
+python src/02_process_batch.py --file_list ./batches/batch_079.txt --output_file checkpoints/batch_079.parquet --checkpoint_every 100
+python src/02_process_batch.py --file_list ./batches/batch_080.txt --output_file checkpoints/batch_080.parquet --checkpoint_every 100
+python src/02_process_batch.py --file_list ./batches/batch_081.txt --output_file checkpoints/batch_081.parquet --checkpoint_every 100
+python src/02_process_batch.py --file_list ./batches/batch_082.txt --output_file checkpoints/batch_082.parquet --checkpoint_every 100
+python src/02_process_batch.py --file_list ./batches/batch_083.txt --output_file checkpoints/batch_083.parquet --checkpoint_every 100
+python src/02_process_batch.py --file_list ./batches/batch_084.txt --output_file checkpoints/batch_084.parquet --checkpoint_every 100
+python src/02_process_batch.py --file_list ./batches/batch_085.txt --output_file checkpoints/batch_085.parquet --checkpoint_every 100
+python src/02_process_batch.py --file_list ./batches/batch_086.txt --output_file checkpoints/batch_086.parquet --checkpoint_every 100
+python src/02_process_batch.py --file_list ./batches/batch_087.txt --output_file checkpoints/batch_087.parquet --checkpoint_every 100
+python src/02_process_batch.py --file_list ./batches/batch_088.txt --output_file checkpoints/batch_088.parquet --checkpoint_every 100
+python src/02_process_batch.py --file_list ./batches/batch_089.txt --output_file checkpoints/batch_089.parquet --checkpoint_every 100

src/04_merge_data.py

@@ -0,0 +1,28 @@
+import pandas as pd
+import glob
+import os
+
+def merge_parquets(input_dir, output_file):
+    # find all .parquet files
+    paths = glob.glob(os.path.join(input_dir, "*.parquet"))
+    if not paths:
+        print(f"No parquet files found in {input_dir}")
+        return
+
+    dfs = []
+    for p in paths:
+        try:
+            df = pd.read_parquet(p)
+            dfs.append(df)
+        except Exception as e:
+            print(f"  ⚠️  Skipping {p}: {e}")
+
+    # concat will union all columns; missing columns → NaN
+    merged = pd.concat(dfs, ignore_index=True, sort=False)
+
+    merged.to_parquet(output_file, index=False)
+    print(f"✅ Merged {len(dfs)} files → {output_file}")
+    print(f"   Total rows: {len(merged)}, Total columns: {len(merged.columns)}")
+
+
+merge_parquets("./checkpoints/","./checkpoints_merged/all_data_merged.parquet")

src/05_sanitize_data.py

@@ -0,0 +1,21 @@
+from molvs import standardize_smiles
+import pandas as pd
+
+file_path = "./checkpoints_merged/all_data_merged.parquet"
+df = pd.read_parquet(file_path)
+smiles_column_key = df.columns[2]
+
+sanatized_rows = []
+std_smile_list=[]
+failed_to_sanatize_row = []
+
+for index, row in df.iterrows():
+    smile = row[smiles_column_key]
+    std_smile = standardize_smiles(smile)
+    std_smile_list.append(std_smile)
+
+df["Smiles_molvs"]=std_smile_list
+col = df.columns[-1]           # name of the last column
+df.insert(3, col, df.pop(col)) # remove it then re‑insert at index 2
+df.drop('SMILES_GDB17', axis=1, inplace=True)
+df.to_parquet("./clean_checkpoints/all_data_merged-cleaned.parquet", index=False)

src/06_calculate_ecfp.py

@@ -0,0 +1,29 @@
+from rdkit import Chem
+from rdkit.Chem import rdFingerprintGenerator
+from rdkit.Chem.Draw import IPythonConsole
+from rdkit import DataStructs
+import rdkit
+import pandas as pd
+
+df=pd.read_parquet("./clean_checkpoints/all_data_merged-cleaned.parquet")
+
+smiles_list=df["Smiles_molvs"]
+ecfp_list=[]
+
+mfpgen = rdFingerprintGenerator.GetMorganGenerator(radius=2,fpSize=2048)
+for i, smile in enumerate(smiles_list):
+    mol = Chem.MolFromSmiles(smile)
+    if mol is None:
+        raise ValueError(f"Invalid SMILES: {smile}, at item", {i})
+    else:
+        fp = mfpgen.GetFingerprint(mol)
+        arr = np.zeros((2048,), dtype=int)            # make empty array
+        DataStructs.ConvertToNumpyArray(fp, arr)      # fill it
+        ecfp_list.append(arr)
+
+df["Ecfp_4"]=ecfp_list
+col = df.columns[-1]          
+df.insert(3, col, df.pop(col)) 
+
+output_file="./ecfp/all_data_merged-cleaned-ecfp4.parquet"
+df.to_parquet(output_file, index=False)

src/07_calculate_properties.py

@@ -0,0 +1,56 @@
+import numpy as np
+import pandas as pd
+from rdkit import Chem
+from rdkit.Chem import Descriptors
+
+
+mw_list=[]
+logp_list=[]
+TPSA_list=[]
+HBD_list=[]
+HBA_list=[]
+rot_bond_list=[]
+ring_count_list=[]
+frac_sp3_list=[]
+
+df=pd.read_parquet("./ecfp/all_data_merged-cleaned-ecfp4.parquet")
+# Loop over SMILES
+
+for smi in df["Smiles_molvs"]:
+    mol = Chem.MolFromSmiles(smi)
+    if mol is None:
+        # if parsing fails, append NaN
+        mw_list.append(np.nan)
+        logp_list.append(np.nan)
+        TPSA_list.append(np.nan)
+        HBD_list.append(np.nan)
+        HBA_list.append(np.nan)
+        rot_bond_list.append(np.nan)
+        ring_count_list.append(np.nan)
+        frac_sp3_list.append(np.nan)
+        print(smi,"this smile failed")
+        continue
+
+    # compute and append
+    mw_list.append(      Descriptors.MolWt(mol)              )
+    logp_list.append(    Descriptors.MolLogP(mol)            )
+    TPSA_list.append(    Descriptors.TPSA(mol)               )
+    HBD_list.append(     Descriptors.NumHDonors(mol)         )
+    HBA_list.append(     Descriptors.NumHAcceptors(mol)      )
+    rot_bond_list.append(Descriptors.NumRotatableBonds(mol)  )
+    ring_count_list.append(Descriptors.RingCount(mol)        )
+    frac_sp3_list.append(Descriptors.FractionCSP3(mol)       )
+
+# Attach back to DataFrame
+df["MolWt"]             = mw_list
+df["ClogP"]             = logp_list
+df["TPSA"]              = TPSA_list
+df["HBD"]               = HBD_list
+df["HBA"]               = HBA_list
+df["NumRotatableBonds"] = rot_bond_list
+df["RingCount"]         = ring_count_list
+df["FractionCSP3"]      = frac_sp3_list
+
+output_file="./ecfp_and_properties/all_data_merged-cleaned-ecfp4-properties.parquet"
+
+df.to_parquet(output_file, index=False)

src/08_organize_columns.py

@@ -0,0 +1,20 @@
+import pandas as pd
+
+df=pd.read_parquet("./ecfp_and_properties/all_data_merged-cleaned-ecfp4-properties.parquet")
+
+new_desc = [
+    "MolWt","ClogP","TPSA","HBD","HBA",
+    "NumRotatableBonds","RingCount","FractionCSP3"
+]
+
+base_cols = ["n_atoms","ID","Smiles_molvs","Ecfp_4"]
+
+other_cols = [c for c in df.columns if c not in base_cols + new_desc]
+
+new_order = base_cols + new_desc + other_cols
+
+df = df[new_order]
+
+output_file="./ecfp_and_properties/all_data_merged-cleaned-ecfp4-properties-sorted-columns.parquet"
+
+df.to_parquet(output_file, index=False)

src/09_unpack_ecfp4.py

@@ -0,0 +1,38 @@
+import pandas as pd
+def expand_array_column(df: pd.DataFrame, column_name: str, prefix: str = None) -> pd.DataFrame:
+    """
+    Expand a column of sequence-like values into multiple scalar columns.
+    
+    Parameters:
+    - df: pandas DataFrame with a column of list/array-like entries.
+    - column_name: name of the column to expand.
+    - prefix: optional prefix for new columns; defaults to column_name.
+    
+    Returns:
+    - A new DataFrame with the original column dropped and new columns added.
+    """
+    # Extract the list/array values
+    sequences = df[column_name].tolist()
+    if not sequences:
+        raise ValueError(f"Column '{column_name}' is empty.")
+    
+    # Determine vector length
+    vec_length = len(sequences[0])
+    # Use provided prefix or fallback to column name
+    prefix = prefix or column_name
+    new_column_names = [f"{prefix}_{i}" for i in range(vec_length)]
+    
+    # Build expanded DataFrame
+    expanded_df = pd.DataFrame(sequences, index=df.index, columns=new_column_names)
+    
+    # Drop original column and concatenate
+    df_dropped = df.drop(columns=[column_name])
+    result_df = pd.concat([df_dropped, expanded_df], axis=1)
+    
+    return result_df
+
+df=pd.read_parquet("./ecfp_and_properties/all_data_merged-cleaned-ecfp4-properties-sorted-columns.parquet")
+result_df=expand_array_column(df,"Ecfp_4","ECFP")
+
+output_file="./ecfp_and_properties/all_data_merged-cleaned-ecfp4-properties-sorted-columns-expanded-ecfp.parquet"
+result_df.to_parquet(output_file, index=False)

src/10_finalize_ml_data.py

@@ -0,0 +1,15 @@
+import pandas as pd
+
+df=pd.read_parquet("./ecfp_and_properties/all_data_merged-cleaned-ecfp4-properties-sorted-columns-expanded-ecfp.parquet")
+
+ID_col     = df.keys()[1]
+chemical_values= df.keys()[3:11]
+thermo_values = df.columns[11:26]
+ecfp_cols   = df.columns[174:]
+
+
+selected_cols = [ID_col] + list(chemical_values) + list(thermo_values)+ list(ecfp_cols)
+
+# slice your DataFrame
+combined_df = df[selected_cols]
+combined_df.to_parquet("./ml_input_data/QML9.parquet") #! final_file

src/11_split_ml_data.py

@@ -0,0 +1,19 @@
+import pandas as pd
+from sklearn.model_selection import train_test_split
+
+
+combined_df = pd.read_parquet("./ml_input_data/QML9.parquet")
+
+
+train_df, temp_df = train_test_split(combined_df, test_size=0.20, random_state=42)
+
+# 2) split temp into 50/50 → 10% each of original
+valid_df, test_df = train_test_split(temp_df, test_size=0.50, random_state=42)
+
+# confirm sizes
+print(f"Train: {len(train_df)/len(df):.2%}")
+print(f"Valid: {len(valid_df)/len(df):.2%}")
+print(f"Test:  {len(test_df)/len(df):.2%}")
+train_df.to_parquet("./ml_input_data/train_split.parquet")
+valid_df.to_parquet("./ml_input_data/validation_split.parquet")
+test_df.to_parquet("./ml_input_data/test_split.parquet")

src/12_train_randomforest.py

@@ -0,0 +1,42 @@
+import pandas as pd
+from sklearn.ensemble import RandomForestRegressor
+from sklearn.metrics import mean_squared_error, r2_score
+
+train_df = pd.read_parquet("./ml_input_data/train_split.parquet")
+valid_df=pd.read_parquet("./ml_input_data/validation_split.parquet")
+test_df=pd.read_parquet("./ml_input_data/test_split.parquet")
+
+chem_properties=list(train_df.keys()[1:9])
+ecfp=list(train_df.keys()[24:])
+target="LUMO_HOMO_GAP"
+feature_cols = chem_properties + ecfp
+
+
+X_train, y_train = train_df[feature_cols], train_df[target]
+X_valid, y_valid = valid_df[feature_cols], valid_df[target]
+X_test,  y_test  = test_df[feature_cols],  test_df[target]
+
+# train simple base line model
+rf_10 = RandomForestRegressor(
+    n_estimators=10,
+    max_depth=None,
+    random_state=42,
+)
+rf_10.fit(X_train, y_train)
+y_pred = rf_10.predict(X_test)
+mse = mean_squared_error(y_test, y_pred)
+r2  = r2_score(y_test,  y_pred)
+
+print(f"Test  MSE: {mse:.4f}")
+print(f"Test  R² : {r2:.4f}") 
+
+#test a subset of molecules with the highest 10% energy gaps
+all_data="./ml_input_data/all_data_no_split.parquet"
+df=pd.read_parquet(all_data)
+top_10_percent = df.sort_values(by='LUMO_HOMO_GAP', ascending=False).head(int(0.1 * len(df)))
+top_10_percent_X_test,  top_10_percent_y_test  = top_10_percent[feature_cols],  top_10_percent[target]
+top_10_percent_y_pred = rf_10.predict(top_10_percent_X_test)
+mse = mean_squared_error(top_10_percent_y_test, top_10_percent_y_pred)
+r2  = r2_score(top_10_percent_y_test,  top_10_percent_y_pred)
+print(f"Top 10% Test  MSE: {mse:.4f}")
+print(f"Top 10% Test  R² : {r2:.4f}")