arXiv subjects

Yoshitaka Inoue

Publications and source records attributed to Yoshitaka Inoue.

8 recordsLinked to original sources

arXiv2026

PerturbRx: Learning Treatment-Conditioned Latent Transitions for Patient Drug Response Prediction

Scarce data and tumor heterogeneity limit patient-level cancer treatment-response prediction. Existing approaches predict response from pretreatment molecular profiles and drug representations, without explicitly modeling the molecular changes expected under treatment. We propose PerturbRx, a treatment-conditioned representation learning framework that learns intervention-induced latent transitions and uses them as patient-drug response features. PerturbRx trains a drug- and dose-conditioned transition predictor from context-matched but cell-unpaired control and treated single-cell populations, then freezes and transfers the predictor to pretreatment patient profiles without requiring post-treatment measurements. The transition is combined with patient and drug representations to predict response. Across TCGA and patient-derived xenograft benchmarks, PerturbRx achieves the strongest aggregate predictive performance among the evaluated methods. These results support perturbation-pretrained latent transitions as useful representations for patient-level drug-response prediction.

q-bio.QM↗

arXiv2026

GraphPINE: Graph Importance Propagation for Interpretable Drug Response Prediction

Explainability is necessary for many tasks in biomedical research. Recent explainability methods have focused on attention, gradient, and Shapley value. These do not handle data with strong associated prior knowledge and fail to constrain explainability results based on known relationships between predictive features. We propose GraphPINE, a graph neural network (GNN) architecture leveraging domain-specific prior knowledge to initialize node importance optimized during training for drug response prediction. Typically, a manual post-prediction step examines literature (i.e., prior knowledge) to understand returned predictive features. While node importance can be obtained for gradient and attention after prediction, node importance from these methods lacks complementary prior knowledge; GraphPINE seeks to overcome this limitation. GraphPINE differs from other GNN gating methods by utilizing an LSTM-like sequential format. We introduce an importance propagation layer that unifies 1) updates for feature matrix and node importance and 2) uses GNN-based graph propagation of feature values. This initialization and updating mechanism allows for informed feature learning and improved graph representation. We apply GraphPINE to cancer drug response prediction using drug screening and gene data collected for over 5,000 gene nodes included in a gene-gene graph with a drug-target interaction (DTI) graph for initial importance. The gene-gene graph and DTIs were obtained from curated sources and weighted by article count discussing relationships between drugs and genes. GraphPINE achieves a PR-AUC of 0.894 and ROC-AUC of 0.796 across 952 drugs. Code is available at https://anonymous.4open.science/r/GraphPINE-40DE.

cs.LG↗

arXiv2026

drGT: Attention-Guided Gene Assessment of Drug Response Utilizing a Drug-Cell-Gene Heterogeneous Network

For translational impact, both accurate drug response prediction and biological plausibility of predictive features are needed. We present drGT, a heterogeneous graph deep learning model over drugs, genes, and cell lines that couples prediction with mechanism-oriented interpretability via attention coefficients (ACs). We assess both predictive generalization (random, unseen-drug, unseen-cell, and zero-shot splits) and biological plausibility (use of text-mined PubMed gene-drug co-mentions and comparison to a structure-based DTI predictor) on GDSC, NCI60, and CTRP datasets. Across benchmarks, drGT consistently delivers top regression performance while maintaining competitive classification accuracy for drug sensitivity. Under random 5-fold cross-validation, drGT attains an AUROC of up to 0.945 (3rd overall) and an $R^2$ up to 0.690, outperforming all baselines on regression. In leave-one-out tests for unseen cell lines and drugs, drGT achieves AUROCs of 0.706 and 0.844, and $R^2$ values of 0.692 and 0.022, the only model yielding positive $R^2$ for unseen drugs. In zero-shot prediction, drGT achieves an AUROC of 0.786 and a regression $R^2$ of 0.334, both representing the highest scores among all models. For interpretability, AC-derived drug-gene links recover known biology: among 976 drugs with known DTIs, 36.9% of predicted links match established DTIs, and 63.7% are supported by either PubMed abstracts or a structure-based predictive model. Enrichment analyses of AC-prioritized genes reveal drug-perturbed biological processes, providing pathway-level explanations. drGT advances predictive generalization and mechanism-centered interpretability, offering state-of-the-art regression accuracy and literature-supported biological hypotheses that demonstrate the use of graph learning from heterogeneous input data for biological discovery. Code: https://github.com/sciluna/drGT

cs.LG↗

arXiv2024

Table Transformers for Imputing Textual Attributes

Missing data in tabular dataset is a common issue as the performance of downstream tasks usually depends on the completeness of the training dataset. Previous missing data imputation methods focus on numeric and categorical columns, but we propose a novel end-to-end approach called Table Transformers for Imputing Textual Attributes (TTITA) based on the transformer to impute unstructured textual columns using other columns in the table. We conduct extensive experiments on three datasets, and our approach shows competitive performance outperforming baseline models such as recurrent neural networks and Llama2. The performance improvement is more significant when the target sequence has a longer length. Additionally, we incorporate multi-task learning to simultaneously impute for heterogeneous columns, boosting the performance for text imputation. We also qualitatively compare with ChatGPT for realistic applications.

cs.CL↗

arXiv2024

Protein-Mamba: Biological Mamba Models for Protein Function Prediction

Protein function prediction is a pivotal task in drug discovery, significantly impacting the development of effective and safe therapeutics. Traditional machine learning models often struggle with the complexity and variability inherent in predicting protein functions, necessitating more sophisticated approaches. In this work, we introduce Protein-Mamba, a novel two-stage model that leverages both self-supervised learning and fine-tuning to improve protein function prediction. The pre-training stage allows the model to capture general chemical structures and relationships from large, unlabeled datasets, while the fine-tuning stage refines these insights using specific labeled datasets, resulting in superior prediction performance. Our extensive experiments demonstrate that Protein-Mamba achieves competitive performance, compared with a couple of state-of-the-art methods across a range of protein function datasets. This model's ability to effectively utilize both unlabeled and labeled data highlights the potential of self-supervised learning in advancing protein function prediction and offers a promising direction for future research in drug discovery.

cs.LG↗

arXiv2024

DrugAgent: Reliable Multi-Agent Integration of Conflicting Biomedical Evidence for Drug-Target Interaction Assessment

Workflows in drug-target interaction (DTI) assessment require integrating heterogeneous data from predictive models, curated resources, and observations from experimental literature. This evidence can be incomplete or conflicting. DrugAgent is a large language model (LLM)-based multi-agent system focused on DTI evidence integration that integrates outputs from machine learning, knowledge graph, and retrieval-augmented generation (RAG) agents. DrugAgent converts agent outputs into interpretable representations, then summarizes conflict across the evidence. We evaluated DrugAgent on kinase screening data of 900 pairs spanning 178 kinases and 42 inhibitors, and an androgen receptor antagonist screening benchmark. On the kinase dataset, LLM-as-a-Judge evaluation indicated outputs were faithful to input evidence in 98.8% of cases. Biological plausibility of returned summarization was high (scores 3-4 out of 5) across ground-truth classes: 79% of Weak activity labels cases (81% for Moderate/77% Strong); Strong cases received higher scores than Weak/Moderate. Label stability showed 98% agreement across runs. Results on the antagonist benchmark were consistent with the kinase dataset. Retrieved literature provided the greatest benefit when direct drug-target evidence was available, highlighting the importance of evidence availability for RAG-based integration. DrugAgent provides heterogeneous evidence-grounded DTI assessment, complementing standalone DTI prediction. We provide strategies to model agreement, conflict, and uncertainty in biomedical evidence integration. Code: https://github.com/sciluna/DrugAgent.

cs.AI↗

arXiv2024

scVGAE: A ZINB-Based Variational Graph Autoencoder for Single-Cell RNA-Seq Imputation

Single-cell RNA sequencing (scRNA-seq) provides high-resolution measurements of cellular heterogeneity, but sparsity and technical zeros can obscure biological structure and complicate downstream analysis. We present scVGAE, a variational graph autoencoder for scRNA-seq imputation that integrates cell-cell graph propagation, a zero-inflated negative binomial (ZINB) likelihood, and direct expression reconstruction. scVGAE constructs a scalable cell graph using principal component analysis (PCA) followed by $k$-nearest neighbors, and encodes each cell into the parameters of a Gaussian latent distribution using graph convolutional networks (GCNs). A low-dimensional latent representation is obtained through stochastic reparameterization and is decoded both into gene-wise ZINB parameters and into a reconstructed expression matrix. Training jointly optimizes ZINB negative log-likelihood, mean-squared reconstruction error, and Kullback--Leibler divergence regularization. We evaluate scVGAE on 14 real-world scRNA-seq datasets against the original expression data and five established imputation methods: MAGIC, ALRA, DeepImpute, DCA, and GNNImpute. scVGAE achieves the highest mean Adjusted Rand Index (ARI) of 0.4681 and the second-highest mean Adjusted Mutual Information (AMI) of 0.5729 across the 14 datasets. These results demonstrate that a compact variational graph representation can preserve cell-class structure competitively across heterogeneous datasets while simultaneously producing an imputed expression matrix.

q-bio.GN↗

arXiv2020

Impairment of insulin-stimulated glucose utilization is associated with burn-induced insulin resistance in mouse muscle by hyperinsulinemic-isoglycemic clamp

Burn-induced insulin resistance is associated with increased morbidity and mortality; however, the impact of burn injury on tissue-specific insulin sensitivity and its molecular mechanisms with consideration of insulin state remains unknown in rodent models. This study was designed to characterize a burn mouse model with tissue-specific insulin resistance under insulin clamp conditions. C57BL6/J mice were subjected to 30% full-thickness burn injury and underwent the combination of hyperinsulinemic isoglycemicclamp (HIC) and positron emission tomography (PET). Hepatic glucose production (HGP) and peripheral glucose disappearance rate (Rd) were measured at different time points up to 7 days post injury. Burned mice showed a significant fasting hypoglycemia and hypoinsulinemia (P < 0.01) on post-burn day (PBD) 3 and 7 along with significantly higher energy expenditure (P < 0.01). HICon PBD 3 demonstrated that burn injury induced systemic insulin resistance, resulting from a significant decrease in insulin-stimulated Rd (33.0 +/- 10.2 vs 68.3 +/- 5.9 mg/kg/min; P < 0.05). In contrast, HGP of burned and sham mice was comparable both in the basal and clamp period. PET on PBD 3 showed a lower insulin-stimulated 18F-labeled 2-fluoro-2-deoxy-D-glucose uptake in the quadriceps of burned mice compared with sham-burned mice. Gastrocnemius muscle harvested from burned mice on PBD 3 showed decreased insulin-stimulated tyrosine phosphorylation of insulin receptor substrate-1 to 34.7% of that in sham-burn mice by immunoblotting analysis (P < 0.05). These findings suggest that impaired insulin-stimulated Rd in skeletal muscle, not elevated HGP, plays a role in the development of burn-induced insulin resistance in a mouse model.

q-bio.TO↗