Search

Abstract Background Deep learning-based drug-target affinity (DTA) prediction methods have shown impressive performance, despite a high number of training parameters relative to the available data. Previous studies have highlighted the presence of dataset bias by suggesting that models trained solely on protein or ligand structures may perform similarly to those trained on complex structures. However, these studies did not propose solutions and focused solely on analyzing complex structure-based models. Even when ligands are excluded, protein-only models trained on complex structures still incorporate some ligand information at the binding sites. Therefore, it is unclear whether binding affinity can be accurately predicted using only compound or protein features due to potential dataset bias. In this study, we expanded our analysis to comprehensive databases and investigated dataset bias through compound and protein feature-based methods using multilayer perceptron models. We assessed the impact of this bias on current prediction models and proposed the binding affinity similarity explorer (BASE) web service, which provides bias-reduced datasets. Results By analyzing eight binding affinity databases using multilayer perceptron models, we confirmed a bias where the compound-protein binding affinity can be accurately predicted using compound features alone. This bias arises because most compounds show consistent binding affinities due to high sequence or functional similarity among their target proteins. Our Uniform Manifold Approximation and Projection analysis based on compound fingerprints further revealed that low and high variation compounds do not exhibit significant structural differences. This suggests that the primary factor driving the consistent binding affinities is protein similarity rather than compound structure. We addressed this bias by creating datasets with progressively reduced protein similarity between the training and test sets, observing significant changes in model performance. We developed the BASE web service to allow researchers to download and utilize these datasets. Feature importance analysis revealed that previous models heavily relied on protein features. However, using bias-reduced datasets increased the importance of compound and interaction features, enabling a more balanced extraction of key features. Conclusions We propose the BASE web service, providing both the affinity prediction results of existing models and bias-reduced datasets. These resources contribute to the development of generalized and robust predictive models, enhancing the accuracy and reliability of DTA predictions in the drug discovery process. BASE is freely available online at https://synbi2024.kaist.ac.kr/base .

The confidentiality threat against training data has become a significant security problem in neural language models. Recent studies have shown that memorized training data can be extracted by injecting well-chosen prompts into generative language models. While these attacks have achieved remarkable success in the English-based Transformer architecture, it is unclear whether they are still effective in other language domains. This paper studies the effectiveness of attacks against Korean models and the potential for attack improvements that might be beneficial for future defense studies. The contribution of this study is two-fold. First, we perform a membership inference attack against the state-of-the-art Korean GPT model. We found approximate training data with 20% to 90% precision in the top-100 samples and confirmed that the proposed attack technique for naive GPT is valid across the language domains. Second, in this process, we observed that the redundancy of the selected sentences could hardly be detected with the existing attack method. Since the information appearing in a few documents is more likely to be meaningful, it is desirable to increase the uniqueness of the sentences to improve the effectiveness of the attack. Thus, we propose a deduplication strategy to replace the traditional word-level similarity metric with the BPE token level. Our proposed strategy reduces 6% to 22% of the underestimated samples among selected ones, thereby improving precision by up to 7%p. As a result, we show that considering both language- and model-specific characteristics is essential to improve the effectiveness of attack strategies. We also discuss possible mitigations against the MI attacks on the general language models.

ObjectiveThis study examined the association of whole-body composition and segmental bioimpedance variables with cold pattern (CP) in different sexes.MethodsWe assigned 667 older individuals to a CP group (n = 488) and a non-CP group (n = 179) by using an eight-item self-administered questionnaire. Seven body composition variables and three pairs of segmental bioimpedance variables for the upper and lower extremities, which were obtained from a segmental multifrequency bioimpedance analyzer, were employed to investigate their association with CP. Participants’ characteristics were first described. Then we compared the selected body composition and bioimpedance variables between the CP and non-CP groups. Finally, their association with CP was investigated using univariate and multivariate regression analyses. All analyses were performed separately for women and men.ResultsBoth women and men exhibited a comparable mean age in the CP and non-CP groups; however, women with CP had significantly lower blood pressures, whereas men with CP showed a higher proportion of osteoarthritis than those without CP. Compared with the non-CP group, individuals with CP exhibited significantly smaller body sizes indicated by shorter height and smaller weight, lower body mass index, and smaller volume-to-body surface area ratio in both sexes. After controlling for age, height, weight, and other covariates, we found significant reductions in body lean mass such as fat-free mass and body cell mass, basal metabolic rate per unit mass, total body water, and intra-to-extracellular water ratio in the CP group. With regard to segmental bioimpedance analysis, the resistance ratios and phase angles in the upper and lower extremities yield significant associations with CP incidence, as demonstrated by the odds ratio (95% confidence interval) of 1.72 (1.16–2.57), 1.69 (1.18–2.48), 0.60 (0.40–0.89), and 0.57 (0.39–0.82), respectively. However, these results did not emerge in men.ConclusionAbnormal cellular water distribution and deterioration in body cell mass and/or cell strength are associated with CP prevalence, regardless of age, height, weight. These findings are similar in the upper and lower extremities and are more pronounced in women. The abovementioned patterns may be considered effective indicators for identifying CP in the older adult population.

Hyperuniform many-particle systems are characterized by a structure factor S(k) that is precisely zero as |k|→0; and stealthy hyperuniform systems have S(k)=0 for the finite range 0