CluelessNovice
commited on
Commit
•
1a22090
1
Parent(s):
ea88ea2
Update README.md
Browse files
README.md
CHANGED
@@ -5,7 +5,17 @@ viewer: True
|
|
5 |
# Description
|
6 |
Subcellular Localization prediction is a 10-class classification task to predict where a protein locates in the cell, where each input protein *x* is mapped to a label *y* ∈ {0, 1, ..., 9}.
|
7 |
|
8 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
9 |
|
10 |
0: Nucleus
|
11 |
|
@@ -26,30 +36,3 @@ The digital label means:
|
|
26 |
8: Lysosome/Vacuole
|
27 |
|
28 |
9: Peroxisome
|
29 |
-
|
30 |
-
# Splits
|
31 |
-
|
32 |
-
**Structure type:** AF2
|
33 |
-
|
34 |
-
The dataset is from [**DeepLoc: prediction of protein subcellular localization using deep learning**](https://academic.oup.com/bioinformatics/article/33/21/3387/3931857). We employ all proteins (proteins that lack AF2 structures are removed), and split them based on 70% structure similarity (see [ProteinShake](https://github.com/BorgwardtLab/proteinshake/tree/main)), with the number of training, validation and test set shown below:
|
35 |
-
|
36 |
-
- Train: 10414
|
37 |
-
- Valid: 1368
|
38 |
-
- Test: 1368
|
39 |
-
|
40 |
-
# Data format
|
41 |
-
|
42 |
-
We organize all data in LMDB format. The architecture of the databse is like:
|
43 |
-
|
44 |
-
**length:** The number of samples
|
45 |
-
|
46 |
-
**0:**
|
47 |
-
|
48 |
-
- **name:** The UniProt ID of the protein
|
49 |
-
|
50 |
-
- **seq:** The structure-aware sequence
|
51 |
-
- **label:** classification label of the sequence
|
52 |
-
|
53 |
-
**1:**
|
54 |
-
|
55 |
-
**···**
|
|
|
5 |
# Description
|
6 |
Subcellular Localization prediction is a 10-class classification task to predict where a protein locates in the cell, where each input protein *x* is mapped to a label *y* ∈ {0, 1, ..., 9}.
|
7 |
|
8 |
+
# Splits
|
9 |
+
|
10 |
+
**Structure type:** AF2
|
11 |
+
|
12 |
+
The dataset is from [**DeepLoc: prediction of protein subcellular localization using deep learning**](https://academic.oup.com/bioinformatics/article/33/21/3387/3931857). We employ all proteins (proteins that lack AF2 structures are removed), and split them based on 70% structure similarity (see [ProteinShake](https://github.com/BorgwardtLab/proteinshake/tree/main)), with the number of training, validation and test set shown below:
|
13 |
+
|
14 |
+
- Train: 10414
|
15 |
+
- Valid: 1368
|
16 |
+
- Test: 1368
|
17 |
+
|
18 |
+
# Label
|
19 |
|
20 |
0: Nucleus
|
21 |
|
|
|
36 |
8: Lysosome/Vacuole
|
37 |
|
38 |
9: Peroxisome
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|