Chemical-protein Interaction Extraction via Gaussian Probability Distribution and External Biomedical Knowledge
In this repository, we provide the codes of our proposed model.
'pytorch_model.bin' in the Resources and Records can be obtained from pytorch_models.
CPI Extraction
-Resources
-NCBI_BERT_pubmed_uncased_L-12_H-768_A-12
-vocab.txt
-pytorch_model.bin
-bert_config.json
-Records
-record_76.56%
-pytorch_model.bin
-bert_config.json
-eval_results.txt
-test_results.txt
-SourceCode
-BioRE.py
-BioRE_BG.py
-modeling.py
-modeling_BG.py
-file_utils.py
-ProcessedData
-CHEMPROT
-train.tsv
-dev.tsv
-test.tsv
-test_overlapping.tsv
-test_normal.tsv
-DDIExtraction2013
-train.tsv
-dev.tsv
-test.tsv
-test_overlapping.tsv
-test_normal.tsv
- Resources: provide the BERT model pre-trained on PubMed.
- Records: provide a record of our proposed model.
- SourceCode: provide the source codes. BioRE.py performs our proposed model; and BioRE.py performs the 'BERT+Gaussian' model.
- ProcessedData: provide the CHEMPROT and DDIExtraction2013 datasets, including the overlapping and normal instances.
- Ubuntu 16.04
- python 3.6.9
- pytorch-pretrained-bert 0.6.1
- torch 1.1.0
- numpy 1.16.4
- pandas 0.25.0
- scipy 1.3.1
Since the model contains multiple layer, it generally need some time to train. If the users have no time to train model, the saved model in the Records can be loaded to test.
Some examples of execution instructions are listed below.
python BioRE.py \
--task_name cpi \
--do_train \
--do_eval \
--do_predict \
--do_lower_case \
--data_dir /$YourPath/CPI_extraction/ProcessedData/CHEMPROT \
--bert_model /$YourPath/CPI_extraction/Resources/NCBI_BERT_pubmed_uncased_L-12_H-768_A-12 \
--max_seq_length 128 \
--train_batch_size 16 \
--eval_batch_size 8 \
--predict_batch_size 8 \
--learning_rate 2e-5 \
--num_train_epochs 2.0 \
--seed 47 \
--output_dir /$YourOutputPath
python BioRE.py \
--task_name cpi \
--do_eval \
--do_predict \
--do_lower_case \
--data_dir /$YourPath/CPI_extraction/ProcessedData/CHEMPROT \
--bert_model /$YourPath/CPI_extraction/Resources/NCBI_BERT_pubmed_uncased_L-12_H-768_A-12 \
--saved_model /$YourSavedmodelPath \
--max_seq_length 128 \
--train_batch_size 16 \
--eval_batch_size 8 \
--predict_batch_size 8 \
--learning_rate 2e-5 \
--num_train_epochs 2.0 \
--output_dir /$YourOutputPath
python BioRE_BG.py \
--task_name cpi \
--do_train \
--do_eval \
--do_predict \
--do_lower_case \
--data_dir /$YourPath/CPI_extraction/ProcessedData/CHEMPROT \
--bert_model /$YourPath/CPI_extraction/Resources/NCBI_BERT_pubmed_uncased_L-12_H-768_A-12 \
--max_seq_length 128 \
--train_batch_size 16 \
--eval_batch_size 8 \
--predict_batch_size 8 \
--learning_rate 2e-5 \
--num_train_epochs 2.0 \
--seed 47 \
--output_dir /$YourOutputPath
python BioRE_BG.py \
--task_name ddi \
--do_train \
--do_eval \
--do_predict \
--do_lower_case \
--data_dir /$YourPath/CPI_extraction/ProcessedData/DDIExtraction2013 \
--bert_model /$YourPath/CPI_extraction/Resources/NCBI_BERT_pubmed_uncased_L-12_H-768_A-12 \
--max_seq_length 128 \
--train_batch_size 16 \
--eval_batch_size 8 \
--predict_batch_size 8 \
--learning_rate 2e-5 \
--num_train_epochs 3.0 \
--seed 17 \
--output_dir /$YourOutputPath
@article{10.1093/bioinformatics/btaa491,
author = {Sun, Cong and Yang, Zhihao and Su, Leilei and Wang, Lei and Zhang, Yin and Lin, Hongfei and Wang, Jian},
title = "{Chemical-protein Interaction Extraction via Gaussian Probability Distribution and External Biomedical Knowledge}",
journal = {Bioinformatics},
year = {2020},
month = {05},
issn = {1367-4803},
doi = {10.1093/bioinformatics/btaa491},
url = {https://doi.org/10.1093/bioinformatics/btaa491},
note = {btaa491},
}
React
Typescript
Django
Laravel
Facebook
Microsoft
Google
Alibaba
D3
Tencent