haneulpark commited on
Commit
5b5646a
·
verified ·
1 Parent(s): 143abe7

Update README.md

Browse files
Files changed (1) hide show
  1. README.md +11 -8
README.md CHANGED
@@ -2,7 +2,7 @@
2
  version: 1.0.0
3
  license: cc-by-sa-4.0
4
  task_categories:
5
- - tabular-regression
6
  language:
7
  - en
8
  pretty_name: MolData
@@ -42,6 +42,8 @@ dataset_info:
42
  dtype: string
43
  - name: 'Y'
44
  dtype: int64
 
 
45
  splits:
46
  - name: train
47
  num_bytes: 12634275804
@@ -76,8 +78,9 @@ This is a mirror of the [Official Github repo](https://github.com/LumosBio/MolDa
76
 
77
  We utilized the raw data uploaded on [Github](https://github.com/LumosBio/MolData) and performed several preprocessing:
78
  1. Sanitize the molecules using RDKit and MolVS (standardize SMILES format)
79
- 2. Rename the columns
80
- 3. Split the dataset (train, test, validation)
 
81
 
82
  If you would like to try these processes with the original dataset,
83
  please follow the instructions in the [Preprocessing Script.py](address) file located in our MolData repository.
@@ -145,9 +148,9 @@ Split and evaluate the catboost model
145
  split_dataset,
146
  column = "SMILES",
147
  representations = load_representations_from_dicts([{"name": "morgan"}, {"name": "maccs_rdkit"}]))
148
-
149
  model = load_model_from_dict({
150
- "name": "cat_boost_regressor",
151
  "config": {
152
  "x_features": ['SMILES::morgan', 'SMILES::maccs_rdkit'],
153
  "y_features": ['Y']}})
@@ -155,11 +158,11 @@ Split and evaluate the catboost model
155
  model.train(split_featurised_dataset["train"])
156
  preds = model.predict(split_featurised_dataset["test"])
157
 
158
- regression_suite = load_suite("regression")
159
 
160
- scores = regression_suite.compute(
161
  references=split_featurised_dataset["test"]['Y'],
162
- predictions=preds["cat_boost_regressor::Y"])
163
 
164
 
165
  ### Citation
 
2
  version: 1.0.0
3
  license: cc-by-sa-4.0
4
  task_categories:
5
+ - tabular-classification
6
  language:
7
  - en
8
  pretty_name: MolData
 
42
  dtype: string
43
  - name: 'Y'
44
  dtype: int64
45
+ description: >-
46
+ Binary classification (0/1)
47
  splits:
48
  - name: train
49
  num_bytes: 12634275804
 
78
 
79
  We utilized the raw data uploaded on [Github](https://github.com/LumosBio/MolData) and performed several preprocessing:
80
  1. Sanitize the molecules using RDKit and MolVS (standardize SMILES format)
81
+ 2. Formatting (from wide form to long form)
82
+ 3. Rename the columns
83
+ 4. Split the dataset (train, test, validation)
84
 
85
  If you would like to try these processes with the original dataset,
86
  please follow the instructions in the [Preprocessing Script.py](address) file located in our MolData repository.
 
148
  split_dataset,
149
  column = "SMILES",
150
  representations = load_representations_from_dicts([{"name": "morgan"}, {"name": "maccs_rdkit"}]))
151
+
152
  model = load_model_from_dict({
153
+ "name": "cat_boost_classifier",
154
  "config": {
155
  "x_features": ['SMILES::morgan', 'SMILES::maccs_rdkit'],
156
  "y_features": ['Y']}})
 
158
  model.train(split_featurised_dataset["train"])
159
  preds = model.predict(split_featurised_dataset["test"])
160
 
161
+ classification_suite = load_suite("classification")
162
 
163
+ scores = classification_suite.compute(
164
  references=split_featurised_dataset["test"]['Y'],
165
+ predictions=preds["cat_boost_classifier::Y"])
166
 
167
 
168
  ### Citation