Datasets:
Hani Park
commited on
Commit
·
eb1b173
1
Parent(s):
cbf7e4b
Removed invalid SMILES
Browse files
.gitattributes
ADDED
@@ -0,0 +1 @@
|
|
|
|
|
1 |
+
*.parquet filter=lfs diff=lfs merge=lfs -text
|
FSMol/{test-00000-of-00001.parquet → FSMol_test.parquet}
RENAMED
@@ -1,3 +1,3 @@
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
-
oid sha256:
|
3 |
-
size
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:a0273f3d3672e4825d73561e705ad64a6776291a1c7c928494d83315ff5cce20
|
3 |
+
size 1023125
|
FSMol/{train-00000-of-00001.parquet → FSMol_train.parquet}
RENAMED
@@ -1,3 +1,3 @@
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
-
oid sha256:
|
3 |
-
size
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:e06be93c72fe2c32a0726249222bc8181365da11de1d8b1fc7d07d7965e84d6a
|
3 |
+
size 127780673
|
FSMol/{validation-00000-of-00001.parquet → FSMol_valid.parquet}
RENAMED
@@ -1,3 +1,3 @@
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
-
oid sha256:
|
3 |
-
size
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:f983145bcae7048a31cf2e01ed7f3a8e5453ee0d95463ed992e136ac710f2fbd
|
3 |
+
size 428531
|
README.md
CHANGED
@@ -67,6 +67,12 @@ This is a mirror of the [Official Github repo](https://github.com/microsoft/FS-M
|
|
67 |
|
68 |
|
69 |
## Preprocessing
|
|
|
|
|
|
|
|
|
|
|
|
|
70 |
|
71 |
We utilized the raw data uploaded on [Github](https://github.com/microsoft/FS-Mol) and performed several preprocessing:
|
72 |
1. Sanitize the molecules using RDKit and MolVS (standardize SMILES format)
|
@@ -105,15 +111,15 @@ and inspecting the loaded dataset
|
|
105 |
DatasetDict({
|
106 |
train: Dataset({
|
107 |
features: ['SMILES', 'Y', 'Assay_ID', 'RegressionProperty', 'LogRegressionProperty', 'Relation'],
|
108 |
-
num_rows:
|
109 |
})
|
110 |
test: Dataset({
|
111 |
features: ['SMILES', 'Y', 'Assay_ID', 'RegressionProperty', 'LogRegressionProperty', 'Relation'],
|
112 |
-
num_rows:
|
113 |
})
|
114 |
validation: Dataset({
|
115 |
features: ['SMILES', 'Y', 'Assay_ID', 'RegressionProperty', 'LogRegressionProperty', 'Relation'],
|
116 |
-
num_rows:
|
117 |
})
|
118 |
|
119 |
})
|
|
|
67 |
|
68 |
|
69 |
## Preprocessing
|
70 |
+
[Update 2025.08.06 Version 1.1.0]
|
71 |
+
We removed invalid SMILES strings from the dataset, which could not be parsed by RDKit.
|
72 |
+
- Train split: removed 12470 strings from 5038727 strings
|
73 |
+
- Test split: removed 127 strings from 56220 strings
|
74 |
+
- Validation split: removed 18 strings from 19008 strings
|
75 |
+
|
76 |
|
77 |
We utilized the raw data uploaded on [Github](https://github.com/microsoft/FS-Mol) and performed several preprocessing:
|
78 |
1. Sanitize the molecules using RDKit and MolVS (standardize SMILES format)
|
|
|
111 |
DatasetDict({
|
112 |
train: Dataset({
|
113 |
features: ['SMILES', 'Y', 'Assay_ID', 'RegressionProperty', 'LogRegressionProperty', 'Relation'],
|
114 |
+
num_rows: 5026257
|
115 |
})
|
116 |
test: Dataset({
|
117 |
features: ['SMILES', 'Y', 'Assay_ID', 'RegressionProperty', 'LogRegressionProperty', 'Relation'],
|
118 |
+
num_rows: 56093
|
119 |
})
|
120 |
validation: Dataset({
|
121 |
features: ['SMILES', 'Y', 'Assay_ID', 'RegressionProperty', 'LogRegressionProperty', 'Relation'],
|
122 |
+
num_rows: 18990
|
123 |
})
|
124 |
|
125 |
})
|