Update README.md
Browse files
README.md
CHANGED
@@ -1,8 +1,34 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
# ProtBert-BFD finetuned on Rosetta 20AA dataset
|
2 |
|
3 |
This model is finetuned to predict Rosetta fold energy using a dataset of 100k 20AA sequences.
|
4 |
|
5 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
6 |
It was trained on protein sequences using a masked language modeling (MLM) objective. It was introduced in
|
7 |
[this paper](https://doi.org/10.1101/2020.07.12.199554) and first released in
|
8 |
[this repository](https://github.com/agemagician/ProtTrans).
|
|
|
|
|
|
1 |
+
---
|
2 |
+
language: protein
|
3 |
+
tags:
|
4 |
+
- protein language model
|
5 |
+
datasets:
|
6 |
+
- BFD
|
7 |
+
- Custom Rosetta
|
8 |
+
---
|
9 |
+
|
10 |
# ProtBert-BFD finetuned on Rosetta 20AA dataset
|
11 |
|
12 |
This model is finetuned to predict Rosetta fold energy using a dataset of 100k 20AA sequences.
|
13 |
|
14 |
+
Current model in this repo: `prot_bert_bfd-finetuned-032722_1752`
|
15 |
+
|
16 |
+
## Performance
|
17 |
+
|
18 |
+
- 20AA sequences (1k eval set):\
|
19 |
+
Metrics: 'mae': 0.090115, 'r2': 0.991208, 'mse': 0.013034, 'rmse': 0.114165
|
20 |
+
|
21 |
+
- 40AA sequences (10k eval set):\
|
22 |
+
Metrics: 'mae': 0.537456, 'r2': 0.659122, 'mse': 0.448607, 'rmse': 0.669781
|
23 |
+
|
24 |
+
- 60AA sequences (10k eval set):\
|
25 |
+
Metrics: 'mae': 0.629267, 'r2': 0.506747, 'mse': 0.622476, 'rmse': 0.788972
|
26 |
+
|
27 |
+
|
28 |
+
## `prot_bert_bfd` from ProtTrans
|
29 |
+
The starting pretrained model is from ProtTrans, trained on 2.1 billion proteins from BFD.
|
30 |
It was trained on protein sequences using a masked language modeling (MLM) objective. It was introduced in
|
31 |
[this paper](https://doi.org/10.1101/2020.07.12.199554) and first released in
|
32 |
[this repository](https://github.com/agemagician/ProtTrans).
|
33 |
+
|
34 |
+
> Created by [Ladislav Rampasek](https://rampasek.github.io)
|