This article offers young scientists a comprehensive guide for selecting a mentor, emphasizing that mentorship is critical for both individual career success and broader scientific innovation. The author shares personal insights, stressing that advisor selection should prioritize scientific ability and mentorship skills over specific research topics. It provides actionable advice on evaluating potential advisors' research output, funding, and the experiences of their current and former trainees. The piece also proposes the development of an "M-index" to measure mentoring quality and advocates for universities and funding bodies to better recognize and reward excellent mentorship alongside scientific achievement.

References:

• Barres B A. How to pick a graduate advisor[J]. Neuron, 2013, 80(2): 275-279.

The podcast offers expert advice on crafting high-quality scientific papers that are both publishable and engaging. It emphasizes the importance of clarity, logical structure, and a compelling narrative to effectively communicate scientific findings. Authors are encouraged to focus on a central message, avoid jargon, and present their work confidently while grounding conclusions in evidence. The sources also highlight the value of making research accessible to a wide audience, including non-specialists, to increase its impact and readership.

This paper introduces a novel approach for integrating 3D imaging with nuclear segmentation to thoroughly analyze cellular neighborhoods in human tissues. The authors highlight the limitations of traditional 2D histology, which loses crucial spatial information, and present their methodology as a way to overcome these challenges. Utilizing the CODA platform, their workflow enables high-resolution 3D reconstructions of tissue architecture and the detailed examination of nuclear morphological features, as demonstrated in a human pancreatic tissue sample. This advanced technique facilitates the spatial characterization of immune cell distribution and offers a more comprehensive understanding of tissue and nuclear structures relevant to disease progression. The study concludes by emphasizing the potential of this 3D nuclear segmentation approach to become a gold standard in digital pathology, despite current computational demands.

References:

• Forjaz A, Kramer D, Shen Y, et al. Integration of nuclear morphology and 3D imaging to profile cellular neighborhoods[J]. biorxiv, 2025: 2025.03. 31.646356.

This paper presents a scientific study investigating the early stages of pancreatic cancer development, focusing on the role of p53 loss and the emergence of progenitor-like cell states. The research utilizes mouse models and single-cell transcriptomics to analyze changes in cell populations, gene expression, and the tumor microenvironment. Specifically, it examines how oncogenic KRAS signaling and p53 inactivation influence premalignant cells and their surroundings, identifying a progenitor niche as a key battleground for malignant transformation. The authors also discuss the methodology employed, including tissue processing, single-cell RNA sequencing, and spatial transcriptomics, alongside references to related work.

References:

• Reyes J, Del Priore I, Chaikovsky A C, et al. Oncogenic and tumor-suppressive forces converge on a progenitor-orchestrated niche to shape early tumorigenesis[J]. bioRxiv, 2025: 2025.06. 10.656791.

The paper introduces CellFM, a novel large-scale foundation model specifically designed for single-cell transcriptomics analysis, built upon a dataset of 100 million human cells. This advanced model, featuring 800 million parameters, aims to overcome the limitations of existing single-cell models by offering enhanced capabilities in tasks such as cell type annotation, perturbation prediction, and gene function prediction. CellFM leverages a modified RetNet architecture, dubbed ERetNet, for improved training efficiency and performance, alongside a Low-Rank Adaptive (LoRA) module to optimize fine-tuning. Extensive experiments demonstrate CellFM's superior accuracy and robustness across various biological applications, showcasing its potential to significantly advance the understanding of cellular states and gene interactions.

References:

• Zeng Y, Xie J, Shangguan N, et al. CellFM: a large-scale foundation model pre-trained on transcriptomics of 100 million human cells[J]. Nature Communications, 2025, 16(1): 4679.

This paper introduces the Asian Immune Diversity Atlas (AIDA), a comprehensive single-cell RNA sequencing reference atlas of human immune cells collected from diverse populations across five Asian countries. The research explores the impact of human diversity, including self-reported ethnicity, genetic ancestry, sex, and age, on the molecular and cellular characteristics of immune cells. Key findings highlight differential abundances of cell neighborhoods and cell populations across various groups, as well as the discovery of population-specific functional variants (eQTLs) that contextualize disease-associated genetic loci. Ultimately, AIDA aims to advance precision medicine by providing a crucial resource for understanding human immune diversity and its implications for disease risk and pathogenesis, especially in underrepresented Asian populations.

References:

• Kock K H, Han K Y, Ando Y, et al. Asian diversity in human immune cells[J]. Cell, 2025, 188(8): 2288-2306. e24.

This comprehensive resource paper presents a multi-omic atlas of human bone marrow, combining single-cell RNA sequencing (scRNA-seq) and co-detection by indexing (CODEX) imaging to map cellular composition, spatial architecture, and cell-cell communication. The study reveals heterogeneity within mesenchymal stromal cells (MSCs), identifying distinct subsets like Adipo- and THY1+ MSCs as primary producers of hematopoietic support factors such as CXCL12, KITLG, CSF1, and IL-7. It further characterizes the spatial organization of hematopoietic stem and progenitor cells (HSPCs), showing a peri-adipocytic localization and an association of early myelopoiesis with a relatively oxygenated arterio-endosteal niche. Finally, the atlas is applied to acute myeloid leukemia (AML), demonstrating stromal expansion and altered cellular neighborhoods in disease, offering a valuable reference for future research into bone marrow niches in health and pathology.

References:

• Bandyopadhyay S, Duffy M P, Ahn K J, et al. Mapping the cellular biogeography of human bone marrow niches using single-cell transcriptomics and proteomic imaging[J]. Cell, 2024, 187(12): 3120-3140. e29.

This article introduces GeneAgent, an advanced AI agent designed to improve gene-set analysis by reducing factual errors, known as hallucinations, often produced by large language models (LLMs). GeneAgent achieves this by autonomously interacting with various biological databases to verify its own generated outputs, ensuring greater accuracy in functional descriptions of gene sets. The research demonstrates that GeneAgent significantly outperforms conventional LLMs like GPT-4 in generating relevant and comprehensive biological process names and explanations. The system's effectiveness is attributed to its self-verification pipeline and its ability to access and leverage expert-curated domain-specific databases.

References:

• Wang Z, Jin Q, Wei C H, et al. GeneAgent: self-verification language agent for gene-set analysis using domain databases[J]. Nature Methods, 2025: 1-9.

This article details a comprehensive study on the progression of borderline ovarian tumors to metastatic low-grade serous cancer. Researchers utilized spatial proteo-transcriptomic profiling, integrating both proteomics and transcriptomics at a cellular level, to map the molecular changes during this transition. The study identifies micropapillary borderline tumors as an intermediate stage and pinpoints specific drivers of cancer transformation. Ultimately, the authors present mechanistic evidence supporting new combination treatments for this chemo-resistant disease, validating potential therapeutic targets through in vitro and in vivo experiments.

References:

• Schweizer L, Kenny H A, Krishnan R, et al. Spatial proteo-transcriptomic profiling reveals the molecular landscape of borderline ovarian tumors and their invasive progression[J]. Cancer Cell, 2025.

The paper introduces CellForge, an innovative agentic system designed for autonomously creating virtual cell models to predict cellular responses to various biological perturbations. This system employs a multi-agent framework, where specialized AI agents collaborate across three core modules: Task Analysis (characterizing datasets and retrieving literature), Method Design (developing optimized modeling strategies through expert discussion), and Experiment Execution (generating and validating executable code). CELLFORGE consistently outperforms existing methods in diverse perturbation prediction tasks, demonstrating significant improvements in accuracy and correlation across different data modalities. Its strength lies in its ability to adaptively design model architectures by integrating biological knowledge and iteratively refining solutions through graph-based expert discussions among agents, rather than relying on predefined heuristics.

References:

• Tang X, Yu Z, Chen J, et al. CellForge: Agentic Design of Virtual Cell Models[J]. arXiv preprint arXiv:2508.02276, 2025.