Yuanfang Guan

We created an interpretable end-to-end neural network (ScoreNet) which can be used to develop an easy-to-implement scoring system for risk stratification. We applied it to data from 2711 patients with advanced melanoma across 5 trials that supported new drug approvals and developed a scoring system for overall survival (OS) and progression-free survival (PFS). The scoring system was then validated in two separate trials. This score is a weighted summation based on eight baseline features (sum of longest diameter for all target lesions, lactate dehydrogenase, systemic immune inflammation index, hemoglobin, ECOG performance status, liver metastasis, pulse rate, and PD-L1 immunohistochemistry status). Using the score, patients can be classified into low, moderate, or high-risk categories of reduced OS and PFS. Risk stratification was demonstrated for patients treated with PD-1/PD-L1 inhibitors, chemotherapy, CTLA-4 inhibitors, BRAF or MEK inhibitors, and their combinations. ScoreNet-based risk stratification can be a new tool for clinical research/practice.

Mapping the olfactory factors directly to perceived smells has broad implications for neuroscience, chemistry, and medicine. Since the discovery of olfactory genes in 1991, the completion of the olfactory code has been hindered by two obstacles: the unknown combinatorial principles by which ~400 receptors collectively encode thousands of perceptually distinct odors, and the absence of a complete functional map from the olfactory receptors to conscious perception. We investigated this problem by using the binding energy scores of odorants with the olfactory receptors. We first showed that using only docking scores, we can predict the smell percepts with an accuracy similar to a full set of chemical fingerprints, combining the two resulted in even better performance. This supports there is a direct relationship between olfactory receptors and specific smells. Next, we turned on the olfactory receptors one by one by iterative training and simulation, and produced corresponding specific perception profiles for each olfactory receptor. The generated matrix is sparse, with only 1-2 smell types activated for each olfactory receptors. Despite the limitation of the size of the training data, it suggests the possibility of a low-dimensional combinatorial principle underlying thousands of smells that humans can perceive. We confirmed the prediction by a list of well-known olfactory receptors. We applied the model trained on single chemicals to mixtures, confirming competitive binding was the driving force for smell specificity. This strong performance, surprisingly, is established on the simplest modeling of the binding scores on hundreds of chemicals, and we believe the mapping can become more accurate if more complicated structural modeling techniques and more data are used.

Gene regulatory networks (GRNs) are essential for understanding how genes coordinate cellular processes. Large-scale single-cell perturbation studies now offer powerful opportunities for GRN inference, yet many state-of-the-art (SOTA) methods fail to fully use interventional information. We present PSGRN, a top-performing method in the CausalBench Challenge, which integrates interventional and observational single-cell RNA sequencing data using a self-training framework with synthetic gold standards. Across eight datasets and six evaluation metrics, PSGRN consistently outperformed existing approaches. With interventional data, it achieved up to 43% higher Wasserstein distances and the lowest false omission rate in K562 compared with recent SOTA methods. Using experimentally validated regulatory interactions, PSGRN showed up to 30% gains in precision and over 100% gains in recall. These results highlight PSGRN’s versatility and scalability, establishing it as a robust tool for GRN inference and biological discovery from single-cell data.

Analysis of Diamond Blackfan anemia syndrome (DBAS) cohorts and animal models has not revealed a potential mechanism for the variable anemia phenotype, a key feature of this disease. Here, we used an established Rpl5Skax23-Jus/+ murine DBAS model to study this dynamic erythropoiesis deficiency. These haploinsufficient mice exhibit variably penetrant craniofacial and cardiac defects mimicking the phenotypes of patients with DBAS bearing RPL5 mutations. We additionally discovered that this specific heterozygous splicing mutation is pathogenic and leads to partial intron retention. By examining the transcriptome of fetal liver erythroid progenitors at embryonic day 12.5 (E12.5), we demonstrate downregulation of erythroid differentiation pathways consistent with the DBAS phenotype. We also identified dysregulated transcription of lipid metabolism genes with significant reduction in Scd1 expression in the subset of E12.5 mutant embryos at risk for erythroid failure. Stearoyl–coenzyme A desaturase 1, a key enzyme that converts saturated to monounsaturated fatty acids, has not been previously linked to erythropoiesis or DBAS. When anemia was induced in adult Rpl5Skax23-Jus/+ mice, mutant mice exhibited delayed erythroid recovery, whereas pretreatment with an stearoyl–coenzyme A desaturase 1 inhibitor resulted in improved erythropoiesis in both wild-type and mutant mice. This analysis suggests a potential role of lipid metabolism in the variable anemia penetrance in DBAS and highlights a previously unappreciated pathway that requires further study as a potential target for drug development.

Decoding visual stimuli from neural signals is an essential step toward understanding how sensory information is represented in the brain. While most existing approaches reconstruct visual stimuli from human functional magnetic resonance imaging (fMRI), utilizing calcium imaging in mice opens the door to single‐neuron‐level insights into non‐primate visual systems with distinct spectral sensitivities. Here, we present Sensorium‐Viz, a diffusion‐based framework specifically designed for decoding activity in the mouse primary visual cortex. The model is among the first to reliably reconstruct complex, high‐resolution images from previously unseen single‐neuron responses. At its core, Sensorium‐Viz introduces two key advances for neuron‐to‐image decoding: a synthetic‐response augmentation strategy that improves reconstruction performance by more than 30% while enabling cross‐mouse generalization through fine‐tuning, and an architectural design that integrates a Diffusion Transformer (DiT) with a spatial neuron‐embedding module, thereby achieving up to a 10.65% performance gain over leading fMRI‐based reconstruction methods across pixel‐ and content‐level benchmarks. Analysis of the neural responses and corresponding reconstructions reveals that neurons sensitive to low‐level visual features form the primary basis of V1's representation of external stimuli. These findings establish Sensorium‐Viz as a biologically grounded and technically robust tool for vision decoding.

Gene networks (GNs) encode diverse molecular relationships and are central to interpreting cellular function and disease. The heterogeneity of interaction types has led to computational methods specialized for particular network contexts. Large language models (LLMs) offer a unified, language-based formulation of GN inference by leveraging biological knowledge from large-scale text corpora, yet their effectiveness remains sensitive to prompt design. Here, we introduce Gene-Relation Adaptive Soft Prompt (GRASP), a parameter-efficient and trainable framework that conditions inference on each gene pair through only three virtual tokens. Using factorized gene-specific and relation-aware components, GRASP learns to map each pair’s biological context into compact soft prompts that combine pair-specific signals with shared interaction patterns. Across diverse GN inference tasks, GRASP consistently outperforms alternative prompting strategies. It also shows a stronger ability to recover unannotated interactions from synthetic negative sets, suggesting its capacity to identify biologically meaningful relationships beyond existing databases. Together, these results establish GRASP as a scalable and generalizable prompting framework for LLM-based GN inference.

Artificial intelligence (AI) is driving innovation in clinical pharmacology and translational science with tools to advance drug development, clinical trials, and patient care. This review summarizes the key takeaways from the AI preconference at the American Society for Clinical Pharmacology and Therapeutics (ASCPT) 2024 Annual Meeting in Colorado Springs, where experts from academia, industry, and regulatory bodies discussed how AI is streamlining drug discovery, dosing strategies, outcome assessment, and patient care. The theme of the preconference was centered around how AI can empower clinical pharmacologists and translational researchers to make informed decisions and translate research findings into practice. The preconference also looked at the impact of large language models in biomedical research and how these tools are democratizing data analysis and empowering researchers. The application of explainable AI in predicting drug efficacy and safety, and the ethical considerations that should be applied when integrating AI into clinical and biomedical research were also touched upon. By sharing these diverse perspectives and real‐world examples, this review shows how AI can be used in clinical pharmacology and translational science to bring efficiency and accelerate drug discovery and development to address patients' unmet clinical needs.

Single-cell sequencing provides detailed insights into individual cell behaviors within complex systems based on the assumption that each cell is uniquely isolated. However, doublets—where two or more cells are sequenced together—disrupt this assumption and can lead to potential data misinterpretations. Traditional doublet detection methods primarily rely on simulated genomic data, which may be less effective in homogeneous cell populations and can introduce biases from experimental processes. Therefore, we introduce ImageDoubler in this study, an innovative image-based model that identifies doublets and missing samples leveraging the Fluidigm single-cell sequencing image data. Our approach showcases a notable doublet detection efficacy, achieving a rate up to 93.87% and registering a minimum improvement of 33.1% in F1 scores compared to existing genomic-based methods. This advancement highlights the potential of using imaging to glean insight into developing doublet detection algorithms and exposes the limitations inherent in current genomic-based techniques.

Modeling neuron responses to stimuli can shed light on next‐generation technologies such as brain‐chip interfaces. Furthermore, high‐performing models can serve to help formulate hypotheses and reveal the mechanisms underlying neural responses. Here the state‐of‐the‐art computational model is presented for predicting single neuron responses to natural stimuli in the primary visual cortex (V1) of mice. The algorithm incorporates object positions and assembles multiple models with different train‐validation data, resulting in a 15%–30% improvement over the existing models in cross‐subject predictions and ranking first in the SENSORIUM 2022 Challenge, which benchmarks methods for neuron‐specific prediction based on thousands of images. Importantly, The model reveals evidence that the spatial organizations of V1 are conserved across mice. This model will serve as an important noninvasive tool for understanding and utilizing the response patterns of primary visual cortex neurons.