LTEnjoy commited on
Commit
f1f914e
·
verified ·
1 Parent(s): 63ee0f4

Upload 7 files

Browse files
Files changed (8) hide show
  1. .gitattributes +3 -0
  2. README.md +52 -0
  3. test/data.mdb +3 -0
  4. test/lock.mdb +0 -0
  5. train/data.mdb +3 -0
  6. train/lock.mdb +0 -0
  7. valid/data.mdb +3 -0
  8. valid/lock.mdb +0 -0
.gitattributes CHANGED
@@ -53,3 +53,6 @@ saved_model/**/* filter=lfs diff=lfs merge=lfs -text
53
  *.jpg filter=lfs diff=lfs merge=lfs -text
54
  *.jpeg filter=lfs diff=lfs merge=lfs -text
55
  *.webp filter=lfs diff=lfs merge=lfs -text
 
 
 
 
53
  *.jpg filter=lfs diff=lfs merge=lfs -text
54
  *.jpeg filter=lfs diff=lfs merge=lfs -text
55
  *.webp filter=lfs diff=lfs merge=lfs -text
56
+ test/data.mdb filter=lfs diff=lfs merge=lfs -text
57
+ train/data.mdb filter=lfs diff=lfs merge=lfs -text
58
+ valid/data.mdb filter=lfs diff=lfs merge=lfs -text
README.md CHANGED
@@ -1,3 +1,55 @@
1
  ---
2
  license: mit
3
  ---
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
  ---
2
  license: mit
3
  ---
4
+ # Description
5
+ 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}.
6
+
7
+ The digital label means:
8
+
9
+ 0: Nucleus
10
+
11
+ 1: Cytoplasm
12
+
13
+ 2: Extracellular
14
+
15
+ 3: Mitochondrion
16
+
17
+ 4: Cell.membrane
18
+
19
+ 5: Endoplasmic.reticulum
20
+
21
+ 6: Plastid
22
+
23
+ 7: Golgi.apparatus
24
+
25
+ 8: Lysosome/Vacuole
26
+
27
+ 9: Peroxisome
28
+
29
+ # Splits
30
+
31
+ **Structure type:** AF2
32
+
33
+ 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:
34
+
35
+ - Train: 10414
36
+ - Valid: 1368
37
+ - Test: 1368
38
+
39
+ # Data format
40
+
41
+ We organize all data in LMDB format. The architecture of the databse is like:
42
+
43
+ **length:** The number of samples
44
+
45
+ **0:**
46
+
47
+ - **name:** The UniProt ID of the protein
48
+
49
+ - **seq:** The structure-aware sequence
50
+ - **plddt**: pLDDT values at all positions
51
+ - **label:** classification label of the sequence
52
+
53
+ **1:**
54
+
55
+ **···**
test/data.mdb ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:764c627aecebdafc1b7bdd87bb3538b71ea549a0803dbfc336863b771e45eb13
3
+ size 9949184
test/lock.mdb ADDED
Binary file (8.19 kB). View file
 
train/data.mdb ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:c70ee8955bd2e00d43a784e39bc2266d07a1cbbf6b8aae39f3a7f424400a7f2e
3
+ size 125423616
train/lock.mdb ADDED
Binary file (8.19 kB). View file
 
valid/data.mdb ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:174f25cff3aa7e76f3e4178ed31452809fc8b6e4e94b180c79e5559e681ba141
3
+ size 10006528
valid/lock.mdb ADDED
Binary file (8.19 kB). View file