MoleculeRankAI/ranking_engine.py at main · 21buddyboyal25-web/MoleculeRankAI · GitHub

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
import pandas as pd
import numpy as np
import joblib


# -----------------------------------------
# Load ADMET data
# -----------------------------------------

admet = pd.read_csv(
    "data/admet_predictions.csv"
)


# -----------------------------------------
# Load activity model
# -----------------------------------------

model = joblib.load(
    "models/rf_model.pkl"
)


# -----------------------------------------
# Create ADMET score
# -----------------------------------------

def admet_score(row):

    score = 100

    # Lower molecular weight preferred
    if row["Molecular_Weight"] > 500:
        score -= 20

    # LogP range preference
    if row["LogP"] > 5:
        score -= 20

    # Too many hydrogen bond donors
    if row["HBD"] > 5:
        score -= 10

    # Too many acceptors
    if row["HBA"] > 10:
        score -= 10

    # High polarity penalty
    if row["TPSA"] > 140:
        score -= 15

    return max(score, 0)


# -----------------------------------------
# Drug-likeness score
# -----------------------------------------

def drug_likeness(row):

    score = 100

    if row["Molecular_Weight"] > 500:
        score -= 25

    if row["Rotatable_Bonds"] > 10:
        score -= 15

    return max(score, 0)


# -----------------------------------------
# Apply scoring
# -----------------------------------------

admet["ADMET_score"] = admet.apply(
    admet_score,
    axis=1
)


admet["Drug_score"] = admet.apply(
    drug_likeness,
    axis=1
)


# -----------------------------------------
# Simple toxicity proxy
# -----------------------------------------

admet["Toxicity_score"] = (
    100 -
    (admet["LogP"].clip(lower=0) * 10)
).clip(
    lower=0
)


# -----------------------------------------
# Activity placeholder
# -----------------------------------------

# Uses normalized activity estimate placeholder
# Full fingerprint prediction will be connected later

# -----------------------------------------
# Load predicted activity
# -----------------------------------------

activity = pd.read_csv(
    "data/activity_predictions.csv"
)


# Normalize activity prediction to 0-100 score

activity["Activity_score"] = (
    (activity["predicted_activity"] - activity["predicted_activity"].min())
    /
    (activity["predicted_activity"].max() - activity["predicted_activity"].min())
) * 100


# Merge activity score into ADMET table

admet = admet.merge(
    activity[
        [
            "canonical_smiles",
            "Activity_score"
        ]
    ],
    on="canonical_smiles",
    how="inner"
)
# -----------------------------------------
# Synthetic accessibility placeholder
# -----------------------------------------

admet["Synthesis_score"] = 70


# -----------------------------------------
# Final ranking score
# -----------------------------------------

admet["Final_score"] = (

    admet["Activity_score"] * 0.30 +

    admet["ADMET_score"] * 0.25 +

    admet["Toxicity_score"] * 0.20 +

    admet["Drug_score"] * 0.15 +

    admet["Synthesis_score"] * 0.10

)


# -----------------------------------------
# Rank molecules
# -----------------------------------------

ranked = admet.sort_values(
    "Final_score",
    ascending=False
)


ranked.to_csv(
    "data/ranked_molecules.csv",
    index=False
)


print("Ranking complete")
print("Molecules ranked:", len(ranked))

print("\nTop 10 molecules:")
print(
    ranked[
        [
            "canonical_smiles",
            "Final_score"
        ]
    ].head(10)
)