Zhentao Liu
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Previous Organizations / Institutes
The organizations / institutes from the scholar's other articles in the past ten years
Zhentao Liu has a total of 71 co-authors. Here are the top co-authors by number of shared publications:
- Yufei Huang
Cancer Virology Program, UPMC Hillman Cancer Center, Pittsburgh, PA, USA
47 shared publications - Yu-Chiao Chiu
Department of Medicine, University of Pittsburgh School of Medicine , Pittsburgh, PA
45 shared publications - Arun Das
Cancer Virology Program, UPMC Hillman Cancer Center, Pittsburgh, PA, USA
42 shared publications - Chien-Hung Shih
UPMC Hillman Cancer Center, University of Pittsburgh School of Medicine, Pittsburgh, Pennsylvania
41 shared publications - George C. Tseng
Department of Biostatistics, University of Pittsburgh, Pittsburgh, USA
41 shared publications - Hugh Galloway
UPMC Hillman Cancer Center, University of Pittsburgh School of Medicine, Pittsburgh, Pennsylvania
41 shared publications
Latest Publications
- Open Access
- 4 May 2026
- Published by Oxford University Press (OUP) in Briefings in Bioinformatics
- Vol. 27 (3)
ST2HE: enhancing spatial transcriptomics interpretability via virtual staining for histological annotation
High-resolution spatial transcriptomics (HR-ST) technologies offer unprecedented insights into tissue architecture but lack standardized frameworks for histological annotation. We present ST2HE, a cross-platform generative framework that synthesizes virtual hematoxylin and eosin images directly from HR-ST data. ST2HE integrates nuclei morphology and spatial transcript coordinates using a one-step diffusion model, enabling histologically informative image generation across diverse tissue types and HR-ST platforms. Conditional and tissue-independent variants support both known and novel tissue contexts. Evaluations on breast cancer, non-small cell lung cancer, and Kaposi’s sarcoma demonstrate ST2HE’s ability to preserve morphological features and support downstream annotations of tissue histology and phenotype classification. Ablation studies reveal that larger context windows, balanced loss functions, and multi-colored transcript visualization enhance image fidelity. ST2HE bridges molecular and histological domains, enabling interpretable, scalable annotation of HR-ST data and advancing computational pathology.
- 4 May 2026
- Published by Oxford University Press (OUP) in Briefings in Bioinformatics
- Vol. 27 (3)
Open AccessST2HE: enhancing spatial transcriptomics interpretability via virtual staining for histological annotation
High-resolution spatial transcriptomics (HR-ST) technologies offer unprecedented insights into tissue architecture but lack standardized frameworks for histological annotation. We present ST2HE, a cross-platform generative framework that synthesizes virtual hematoxylin and eosin images directly from HR-ST data. ST2HE integrates nuclei morphology and spatial transcript coordinates using a one-step diffusion model, enabling histologically informative image generation across diverse tissue types and HR-ST platforms. Conditional and tissue-independent variants support both known and novel tissue contexts. Evaluations on breast cancer, non-small cell lung cancer, and Kaposi’s sarcoma demonstrate ST2HE’s ability to preserve morphological features and support downstream annotations of tissue histology and phenotype classification. Ablation studies reveal that larger context windows, balanced loss functions, and multi-colored transcript visualization enhance image fidelity. ST2HE bridges molecular and histological domains, enabling interpretable, scalable annotation of HR-ST data and advancing computational pathology.
- Open Access
- 17 December 2025
- Published by Springer Nature in Research Square
A Process-Centric Survey of AI for Scientific Discovery Through the EXHYTE Framework
Large language models (LLMs) and agent systems are increasingly transforming scientific discovery, driving progress across chemistry, biology, materials science, and physics. Yet most existing work and surveys remain fragmented, focusing on isolated tasks such as idea generation or experiment design without addressing how these components fit within the broader discovery process. To bridge this gap, we introduce the EXHYTE cycle, an iterative framework that formalizes scientific discovery as a sequence of Exploration, Hypothesis generation, and Testing. We assembled a corpus of recent studies, distilled recurring strategies that characterize how AI methods contribute to each EXHYTE substage, and organized the literature accordingly to representative strategies and domain-specific advances. This process-centric perspective unifies diverse methodologies under a single structured workflow, identifies substages that are mature versus underexplored, and reveals complementarities that enable closed-loop discovery systems. It also clarifies the evolving division of labor between human researchers and AI systems, offering a roadmap for developing adaptive, autonomous frameworks for AIdriven scientific discovery.An accompanying website with paper summaries and an LLM-powered interactive survey based on EXHYTE is available at https: //webapps.crc.pitt.edu/exhyte/
- 17 December 2025
- Published by Springer Nature in Research Square
Open AccessA Process-Centric Survey of AI for Scientific Discovery Through the EXHYTE Framework
Large language models (LLMs) and agent systems are increasingly transforming scientific discovery, driving progress across chemistry, biology, materials science, and physics. Yet most existing work and surveys remain fragmented, focusing on isolated tasks such as idea generation or experiment design without addressing how these components fit within the broader discovery process. To bridge this gap, we introduce the EXHYTE cycle, an iterative framework that formalizes scientific discovery as a sequence of Exploration, Hypothesis generation, and Testing. We assembled a corpus of recent studies, distilled recurring strategies that characterize how AI methods contribute to each EXHYTE substage, and organized the literature accordingly to representative strategies and domain-specific advances. This process-centric perspective unifies diverse methodologies under a single structured workflow, identifies substages that are mature versus underexplored, and reveals complementarities that enable closed-loop discovery systems. It also clarifies the evolving division of labor between human researchers and AI systems, offering a roadmap for developing adaptive, autonomous frameworks for AIdriven scientific discovery.An accompanying website with paper summaries and an LLM-powered interactive survey based on EXHYTE is available at https: //webapps.crc.pitt.edu/exhyte/
- Open Access
- 13 May 2025
- Published by American Association for Cancer Research (AACR)
Table S7 from Inferring Drug–Gene Relationships in Cancer Using Literature-Augmented Large Language Models
Table S7 from Inferring Drug–Gene Relationships in Cancer Using Literature-Augmented Large Language Models
- 13 May 2025
- Published by American Association for Cancer Research (AACR)
Open AccessTable S7 from Inferring Drug–Gene Relationships in Cancer Using Literature-Augmented Large Language Models
Table S7 from Inferring Drug–Gene Relationships in Cancer Using Literature-Augmented Large Language Models
- Open Access
- 13 May 2025
- Published by American Association for Cancer Research (AACR)
Figure 4 from Inferring Drug–Gene Relationships in Cancer Using Literature-Augmented Large Language Models
Drug–target network in liver cancer. A, Drug–target network in liver cancer. The network was constructed by extracting drug–gene pairs associated with liver cancer from Fig. 3A, in which liver cancer was identified as a primary cancer type among multiple possible types generated through GPT-4o inference. The node size and font size are proportional to the degree (i.e., the number of connections). Drugs and genes discussed in the main text are highlighted in bold. B, Kaplan–Meier curves for progression-free survival of patients with HCC, the major type of liver cancer, treated with sorafenib in the IMBrave150 clinical trial, comparing CTNNB1-mutated (CTNNB1-Mut; n = 19) and CTNNB1 wild-type patients (CTNNB1-WT; n = 26). Statistical significance was assessed by the log-rank test. C, Boxplots showing differential sensitivity to sorafenib between CTNNB1-mutated (CTNNB1-Mut; n = 2) and CTNNB1 wild-type liver cancer cell lines (CTNNB1-WT; n = 18). Drug sensitivity data are represented as the log2 fold change in cell counts following treatment, derived from the Profiling Relative Inhibition Simultaneously in Mixtures (PRISM) drug screening project. Statistical significance was assessed by the one-tailed Mann–Whitney U test. D, Correlation between CTNNB1 genetic dependency (represented by the Chronos gene-effect score from DepMap) and sorafenib sensitivity (PRISM) in liver cancer cell lines (n = 18). A more negative gene-effect score represents a stronger genetic dependency. The names of cell lines exhibiting strong dependency on CTNNB1 (with gene-effect scores < −1) are labeled.
- 13 May 2025
- Published by American Association for Cancer Research (AACR)
Open AccessFigure 4 from Inferring Drug–Gene Relationships in Cancer Using Literature-Augmented Large Language Models
Drug–target network in liver cancer. A, Drug–target network in liver cancer. The network was constructed by extracting drug–gene pairs associated with liver cancer from Fig. 3A, in which liver cancer was identified as a primary cancer type among multiple possible types generated through GPT-4o inference. The node size and font size are proportional to the degree (i.e., the number of connections). Drugs and genes discussed in the main text are highlighted in bold. B, Kaplan–Meier curves for progression-free survival of patients with HCC, the major type of liver cancer, treated with sorafenib in the IMBrave150 clinical trial, comparing CTNNB1-mutated (CTNNB1-Mut; n = 19) and CTNNB1 wild-type patients (CTNNB1-WT; n = 26). Statistical significance was assessed by the log-rank test. C, Boxplots showing differential sensitivity to sorafenib between CTNNB1-mutated (CTNNB1-Mut; n = 2) and CTNNB1 wild-type liver cancer cell lines (CTNNB1-WT; n = 18). Drug sensitivity data are represented as the log2 fold change in cell counts following treatment, derived from the Profiling Relative Inhibition Simultaneously in Mixtures (PRISM) drug screening project. Statistical significance was assessed by the one-tailed Mann–Whitney U test. D, Correlation between CTNNB1 genetic dependency (represented by the Chronos gene-effect score from DepMap) and sorafenib sensitivity (PRISM) in liver cancer cell lines (n = 18). A more negative gene-effect score represents a stronger genetic dependency. The names of cell lines exhibiting strong dependency on CTNNB1 (with gene-effect scores < −1) are labeled.
- Open Access
- 13 May 2025
- Published by American Association for Cancer Research (AACR)
Table S7 from Inferring Drug–Gene Relationships in Cancer Using Literature-Augmented Large Language Models
Table S7 from Inferring Drug–Gene Relationships in Cancer Using Literature-Augmented Large Language Models
- 13 May 2025
- Published by American Association for Cancer Research (AACR)
Open AccessTable S7 from Inferring Drug–Gene Relationships in Cancer Using Literature-Augmented Large Language Models
Table S7 from Inferring Drug–Gene Relationships in Cancer Using Literature-Augmented Large Language Models