Large-scale protein-protein post-translational modification extraction with distant supervision and confidence calibrated BioBERT

Motivation Protein-protein interactions (PPIs) are critical to normal cellular function and are related to many disease pathways. A range of protein functions are mediated and regulated by protein interactions through post-translational modifications (PTM). However, only 4% of PPIs are annotated with PTMs in biological knowledge databases such as IntAct, mainly performed through manual curation, which is neither time- nor cost-effective. Here we aim to facilitate annotation by extracting PPIs along with their pairwise PTM from the literature by using distantly supervised training data using deep learning to aid human curation. Method We use the IntAct PPI database to create a distant supervised dataset annotated with interacting protein pairs, their corresponding PTM type, and associated abstracts from the PubMed database. We train an ensemble of BioBERT models—dubbed PPI-BioBERT-x10—to improve confidence calibration. We extend the use of ensemble average confidence approach with confidence variation to counteract the effects of class imbalance to extract high confidence predictions. Results and conclusion The PPI-BioBERT-x10 model evaluated on the test set resulted in a modest F1-micro 41.3 (P =5 8.1, R = 32.1). However, by combining high confidence and low variation to identify high quality predictions, tuning the predictions for precision, we retained 19% of the test predictions with 100% precision. We evaluated PPI-BioBERT-x10 on 18 million PubMed abstracts and extracted 1.6 million (546507 unique PTM-PPI triplets) PTM-PPI predictions, and filter $$\approx 5700$$ ≈ 5700 (4584 unique) high confidence predictions. Of the 5700, human evaluation on a small randomly sampled subset shows that the precision drops to 33.7% despite confidence calibration and highlights the challenges of generalisability beyond the test set even with confidence calibration. We circumvent the problem by only including predictions associated with multiple papers, improving the precision to 58.8%. In this work, we highlight the benefits and challenges of deep learning-based text mining in practice, and the need for increased emphasis on confidence calibration to facilitate human curation efforts.

Large-Scale Protein-Protein Post-Translational Modification Extraction With Distant Supervision And Confidence Calibrated BioBERT

Motivation: Protein-protein interactions (PPIs) are critical to normal cellular function and are related to many disease pathways. A range of protein functions are mediated and regulated by protein interactions through post-translational modifications (PTM). However, only 4% of PPIs are annotated with PTMs in biological knowledge databases such as IntAct, mainly performed through manual curation, which is neither time- nor cost-effective. Here we aim to facilitate annotation by extracting PPIs along with their pairwise PTM from the literature by using distantly supervised training data using deep learning to aid human curation. We further assessed model generalisation in a real-world scenario, evaluating its performance on a randomly sampled subset of predictions from 18 million PubMed abstracts. Method: We use the IntAct PPI database to create a distant supervised dataset annotated with interacting protein pairs, their corresponding PTM type, and associated abstracts from the PubMed database. We train an ensemble of BioBERT models – dubbed PPI-BioBERT-x10 – to improve confidence calibration. We extend the use of ensemble average confidence approach with confidence variation to counteract the effects of class imbalance to extract high confidence predictions. Results and conclusion: The PPI-BioBERT-x10 model evaluated on the test set resulted in a modest F1-micro 41.3 (P=58.1, R=32.1). However, by combining high confidence and low variation to identify high quality predictions, tuning the predictions for precision, we retained 19% of the test predictions with 100% precision. We evaluated PPI-BioBERT-x10 on 18 million PubMed abstracts and extracted 1.6 million (546507 unique PTM-PPI triplets) PTM-PPI predictions, and filter ≈5,700 (4584 unique) high confidence predictions. Of the 5700, human evaluation on a small randomly sampled subset shows that the precision drops to 33.7% despite confidence calibration and highlights the challenges of generalisability beyond the test set even with confidence calibration. We circumvent the problem by only including predictions associated with multiple papers, improving the precision to 58.8%. In this work, we highlight the benefits and challenges of deep learning-based text mining in practice, and the need for increased emphasis on confidence calibration to facilitate human curation efforts.

Moderators of relative confidence calibration: A meta-analysis of the across-subject relationship between confidence and accuracy

If one friend confidently tells us to buy product A while another friend thinks that product B is better but is not confident, we may go with the advice of our confident friend. Should we? The relationship between people’s confidence and accuracy has been of great interest in many fields, especially in the context of high-stakes situations like eye-witness testimony, but there is still little consensus about how much we should trust someone’s overall level of confidence. Here we examine the across-subject relationship between average accuracy and average confidence in 214 unique datasets from the Confidence Database. This approach allows us to empirically address this issue with unprecedented statistical power and check for the presence of various moderators. We find that the across-subject correlation between average accuracy and average confidence in this sample is R = .22. Importantly, this relationship is much stronger for memory than for perception tasks, as well as for confidence scales with fewer points. These results show that we should take one’s confidence seriously (and perhaps buy product A) and suggest several factors that moderate the relative consistency of how people make confidence judgments.