A Transferability-Based Method for Evaluating the Protein Representation Learning.

Self-supervised pre-trained language models have recently risen as a powerful approach in learning protein representations, showing exceptional effectiveness in various biological tasks, such as drug discovery. Amidst the evolving trend in protein language model development, there is an observable shift towards employing large-scale multimodal and multitask models. However, the predominant reliance on empirical assessments using specific benchmark datasets for evaluating these models raises concerns about the comprehensiveness and efficiency of current evaluation methods. Addressing this gap, our study introduces a novel quantitative approach for estimating the performance of transferring multi-task pre-trained protein representations to downstream tasks. This transferability-based method is designed to quantify the similarities in latent space distributions between pre-trained features and those fine-tuned for downstream tasks. It encompasses a broad spectrum, covering multiple domains and a variety of heterogeneous tasks. To validate this method, we constructed a diverse set of protein-specific pre-training tasks. The resulting protein representations were then evaluated across several downstream biological tasks. Our experimental results demonstrate a robust correlation between the transferability scores obtained using our method and the actual transfer performance observed. This significant correlation highlights the potential of our method as a more comprehensive and efficient tool for evaluating protein representation learning.

A Multimodal Protein Representation Framework for Quantifying Transferability Across Biochemical Downstream Tasks.

Proteins are the building blocks of life, carrying out fundamental functions in biology. In computational biology, an effective protein representation facilitates many important biological quantifications. Most existing protein representation methods are derived from self-supervised language models designed for text analysis. Proteins, however, are more than linear sequences of amino acids. Here, a multimodal deep learning framework for incorporating ≈1 million protein sequence, structure, and functional annotation (MASSA) is proposed. A multitask learning process with five specific pretraining objectives is presented to extract a fine-grained protein-domain feature. Through pretraining, multimodal protein representation achieves state-of-the-art performance in specific downstream tasks such as protein properties (stability and fluorescence), protein-protein interactions (shs27k/shs148k/string/skempi), and protein-ligand interactions (kinase, DUD-E), while achieving competitive results in secondary structure and remote homology tasks. Moreover, a novel optimal-transport-based metric with rich geometry awareness is introduced to quantify the dynamic transferability from the pretrained representation to the related downstream tasks, which provides a panoramic view of the step-by-step learning process. The pairwise distances between these downstream tasks are also calculated, and a strong correlation between the inter-task feature space distributions and adaptability is observed.

Bridging the Gap between Target-Based and Cell-Based Drug Discovery with a Graph Generative Multitask Model.

The development of new drugs is crucial for protecting humans from disease. In the past several decades, target-based screening has been one of the most popular methods for developing new drugs. This method efficiently screens potential inhibitors of a target protein in vitro, but it frequently fails in vivo due to insufficient activity of the selected drugs. There is a need for accurate computational methods to bridge this gap. Here, we present a novel graph multi-task deep learning model to identify compounds with both target inhibitory and cell active (MATIC) properties. On a carefully curated SARS-CoV-2 data set, the proposed MATIC model shows advantages compared with the traditional method in screening effective compounds in vivo. Following this, we investigated the interpretability of the model and discovered that the learned features for target inhibition (in vitro) or cell active (in vivo) tasks are different with molecular property correlations and atom functional attention. Based on these findings, we utilized a Monte Carlo-based reinforcement learning generative model to generate novel multiproperty compounds with both in vitro and in vivo efficacy, thus bridging the gap between target-based and cell-based drug discovery. The tool is freely accessible at https://github.com/SIAT-code/MATIC.

A single-center 14-year follow-up study of the BalMedic® bovine pericardial bioprosthetic valve.

BACKGROUND: The causes of valvular disease in China are complex, with a broad age distribution. For patients with early mechanical valve replacement, the quality of life is affected by postoperative anticoagulation-related complications. Since 2005, we have used bioprosthetic valves to provide more options for patients. In this study, we retrospectively analyzed the 14-year follow-up data of patients undergoing BalMedic bovine pericardial bioprosthetic valve replacement (manufacturer: Beijing Balance Medical Tech Co., Ltd.) to evaluate its intermediate- to long-term clinical effectiveness. METHODS: From 2005 to 2014, 336 BalMedic pericardial bioprosthesis valves were implanted in 299 patients (mean age 53.5 years, 59.86% female) at The First People's Hospital of Yulin. Among followed up 290 discharged patients, 284 underwent aortic valve replacement and mitral valve replacement (AVR group, MVR group) for further grouping analysis, 6 underwent tricuspid valve replacement (TVR). The mean follow-up was 7.7±2.5 years (5 to 14), for a total of 2,196 valve-years, 98.28% of the patients completed follow-up. RESULTS: The perioperative mortality was 3% (9/299). After discharge, 68 patients (23.4%, 68/290) died, and 36 (12.4% 36/290) underwent the second valve replacement. The overall 5- and 10-year survival rates were 89.95% and 72.53%, respectively. For patients undergoing AVR alone, the overall 10-year survival rates were 80.64%, the reoperation-free rates were 92.94%, and the SVD-free rates were 90.95%. For patients undergoing MVR and double valve replacement (DVR group), the 10-year survival rates were 67.21% and 82.90%, the reoperation-free rates were 72.26% and 73.33%, the SVD-free rates were 58.90% and 53.80%, respectively. Subgroup analysis by age showed no significant intergroup difference in overall survival but a significant intergroup difference in reoperation-free and SVD-free rates (P<0.05). CONCLUSIONS: With a similar 10-year overall survival rate as its foreign counterparts, BalMedic bovine pericardial bioprosthesis is reliable choice for both aortic valve and mitral valve. In patients undergoing AVR, the BalMedic valve is superior to the similar foreign counterparts in overall survival, reoperation-free survival, and SVD-free rates. While in MVR or DVR, Chinese patients are younger because of different etiology, postoperative outcomes show non-inferior to those from the foreign counterparts.