Biology subjects

Van Allen, E.

Publications and source records attributed to Van Allen, E..

At least 19 recordsLinked to original sources

bioRxiv2026

Evaluating agentic AI for biological discovery in autonomous and copilot settings

Advances in large language models (LLMs)-based artificial intelligence (AI) agents have improved their ability to execute structured analytical workflows, including standard bioinformatic pipelines for biological discovery. However, computational biology rarely consists of deterministic pipeline execution alone. Biological datasets are heterogeneous and noisy, and meaningful discovery often requires open-ended hypothesis generation and iterative reasoning over multimodal evidence. These challenges are particularly evident in multi-omic studies, where paired molecular modalities and heterogeneous clinical contexts create both opportunities and obstacles for discovery. The extent to which emerging agentic AI systems can support or automate this mode of scientific discovery remains poorly understood. Here, we systematically evaluated the capabilities and limitations of agentic AI for biological discovery using multi-omic single cell datasets spanning 11 cancer types. We developed the Multistep Multimodal Multiomic Agentic (M3A) Framework to support LLM-driven reasoning over persistent multimodal data states and to capture agentic reasoning behavior in autonomous and human-AI copilot settings. Using this framework, we assessed AI agents across complementary tasks, including autonomous cell-type annotation, generation of falsifiable biological hypotheses from gene programs, and copilot experiments testing the effect of human involvement and domain expertise. We found that current AI agents are effective at broad, systemic exploration of complex data, whereas domain experts remain critical for methodological guidance and biological synthesis across analyses. Together, our results delineate the current potential and boundaries of agentic AI in computational biology, and establish a framework for evaluating AI systems designed to support biological discovery.

cancer biology↗

bioRxiv2026

Expanding P-NET, a multi-purpose biologically informed deep learning framework

We present expanded P-NET, a versatile frame-work for deep learning in computational biology based on P-NET, leveraging biological pathways for interpretable predictions. Our framework achieves competitive performance in genomic & transcriptomic prediction tasks. We demonstrate its stability and interpretability compared to traditional machine learning models. P-NET 2.0 incorporates gene and pathway information, providing valuable insights into complex biological processes. The framework is publicly available, enabling its application to various computational biology tasks.

bioinformatics↗

bioRxiv2026

SpatialFusion: A lightweight multimodal foundation model for pathway-informed spatial niche mapping

Foundation models enable knowledge transfer across data modalities and tasks, yet foundation models for spatial biology remain in their early stages, largely centered on encoding single-cell representations in spatial context without fully integrating transcriptomic and morphological information to delineate functional niches. Here we introduce SpatialFusion, a lightweight multimodal foundation model that identifies biologically coherent microenvironments defined by distinct pathway activation patterns rather than spatial proximity alone. SpatialFusion integrates paired histopathology, gene expression, and inferred pathway activity into a unified representation. Compared with two specialist niche-detection methods and four spatial foundation models, SpatialFusion performs competitively and consistently resolves fine-grained spatial niches with unique pathway-level signatures. Applying the model to two Visium HD cohorts uncovered a pre-malignant niche in morphologically normal mucosa adjacent to colorectal tumors and revealed distinct malignant microenvironments in non-small cell lung cancer that were predictive of tumor stage. Overall, SpatialFusion offers a versatile framework for multimodal spatial analysis, enabling the discovery of new morpho-molecular niches with significant biological and clinical relevance.

bioinformatics↗

bioRxiv2025

Conserved Neuronal-like and Secretory Programs Define the Spatial Architecture of Gastroenteropancreatic Neuroendocrine Tumors

Gastroenteropancreatic neuroendocrine tumors (GEP-NETs) are clinically heterogeneous malignancies whose biology and microenvironmental organization remain poorly understood. Here, we integrated single-nucleus multiomic (snRNA-seq and snATAC-seq) and spatial transcriptomic profiling across 38 well-differentiated pancreatic (pNET) and small-intestinal (siNET) tumors to define conserved malignant programs, their regulatory circuits, and spatial niches. We observed two conserved malignant cell programs spanning a continuous transcriptional spectrum: a neuronal-like program (si-cNMF1/p-cNMF1), and a secretory neuroendocrine program (si-cNMF2/p-cNMF2). Matched chromatin accessibility profiles uncovered distinct, tissue-specific regulatory networks, including MAX::MYC and MITF transcription factor binding motifs in siNETs versus ISL1 and TFAP4 in pNETs, indicating organ-specific epigenetic control. Spatial transcriptomic analyses revealed that si/p-cNMF1-high regions localized to high cell density, immune-rich tumor areas, whereas si/p-cNMF2-high regions occupied stromal and vascularized niches and co-occured with fibroblast and endothelial compartments enriched for TGFB1-ITGB1, VEGFA-FLT1, and LAMA2-ITGA1 signaling. Across both tumor types, the cNMF2 program was enriched in metastatic lesions and was enrichedfor pro-fibrotic and pro-angiogenic gene signatures. Thus, GEP-NETs are organized along a conserved neuronal-to-secretory axis defined by distinct epigenetic programs and spatially coupled to specific microenvironmental niches. This framework unifies NET heterogeneity across organ sites and identifies pathway-specific, microenvironment-linked vulnerabilities for therapeutic targeting.

cancer biology↗

bioRxiv2025

Mechanisms of tumor persistence in metastatic melanoma following successful immunotherapy

Tumor dormancy is thought to enable cancer recurrence, but the settings and mechanisms of dormancy in patients are poorly characterized. Both immunogenic tumor mass dormancy, involving continued tumor cell proliferation and balanced immune-mediated cell death, and cell-level quiescence have been observed in preclinical models. Here, we demonstrate the existence of mass dormancy rather than quiescence in long-term stable residual lesions in melanoma patients treated with immune checkpoint inhibitors (ICI). In a large stable lesion subjected to detailed spatial profiling, the proportion of proliferating tumor cells is similar to that in site-matched tumors from patients progressing after ICI. Residual stable lesions from other patients, judged to contain only scar tissue upon clinical pathology review, also contained nests of dividing tumor cells surrounded by active immune cells. These findings demonstrate the presence of viable tumor cells and mass dormancy in persistent stable lesions, a finding with implications for disease monitoring and management.

cancer biology↗

bioRxiv2025

Simulation and empirical evaluation of biologically-informed neural network performance

Biologically-informed neural networks (BiNNs) offer interpretable deep learning models for biological data, but the dataset characteristics required for strong performance remain poorly understood. For instance, we previously developed P-NET, a BiNN with an architecture based on the Reactome pathway database, and applied this model to predict metastatic status of patients with prostate cancer using somatic mutation and copy number information. It seems likely that including additional relevant signal - e.g., germline variation in this context - should improve model performance, but we currently lack a principled approach to assess whether BiNNs will successfully detect this signal. Here, we developed two simulation frameworks to evaluate the factors that influence BiNN performance - including signal type, signal strength, feature sparsity, and sample size - and empirically tested how integrating germline and somatic data affects the models ability to predict prostate cancer metastatic status. Simulations revealed that small sample size, weak signal strength, and especially extreme feature sparsity limit BiNN performance, and that the model preferentially uses linear over nonlinear signal. Empirically, P-NET performed poorly on sparse germline data, and while adding germline to somatic data did not improve prediction, it improved gene prioritization and model interpretation. Broadly, our simulation frameworks enable systematic evaluation of how dataset-level characteristics affect BiNN performance and provide a principled framework for benchmarking novel methods.

bioinformatics↗

bioRxiv2025

Empirical Evaluation of Single-Cell Foundation Models for Predicting Cancer Outcomes

Foundation models pretrained on large-scale single-cell RNA sequencing data present a promising opportunity to advance translational cancer research. However, their utility in clinically relevant, patient-level single-cell applications remains underexplored. Here, we developed an agentic strategy to systematically evaluate twelve emerging single-cell foundation models (scFMs) and three alternative baseline approaches across seven cancer-specific tasks, including cell-type annotation, cancer subtype classification, and treatment response prediction. We assessed model performance under zero-shot, continual training, and fine-tuning conditions, conducting 1,530 supervised model-fitting runs and 200 unsupervised subsample evaluations. We found that while current scFMs excelled at certain analysis tasks, such as tumor microenvironment cell annotation, they offered limited advantages in predicting clinical and biological outcomes of cancer patients compared to simpler baseline models. These insights highlight the critical role of scFM evaluation on biologically and clinically relevant tasks for precision oncology. Beyond identifying current limitations, this assessment reveals principles that can guide future methodological innovation and the use of expanded cancer single-cell cohorts to build more biologically informed and translationally effective scFMs. The resulting agentic framework supports the autonomous discovery of emerging scFMs and facilitates their standardized integration and evaluation across cancer-related tasks.

cancer biology↗

bioRxiv2025

Segregating DNA lesions point to high selective advantage of tumor initiating cells

The complications with identifying cells at the origin of cancer and tracking their early divisions impede studies of cancer initiation. Recently, it was shown that some DNA lesions generated by a pulse of damage-inducing mutagen persist over multiple rounds of replication. Segregation of DNA lesions in the early genealogy of an expanding clone leaves a statistically interpretable footprint of cancer initiating events. Specifically, it allows for estimating the number of cell divisions between the initiating DNA lesion and the most recent common ancestor of the tumor. Here, we analyze footprints of segregating lesions from a previously published experimental mouse system, as well as post-chemotherapy human metastatic tumors and the blood of chemotherapy treated patients. In all contexts, clones tend to start early, usually within the span of 4 cell generations from mutagen exposure. Using a branching process model, we show that fitness advantage of early cancer drivers exceeds 30%, with each early division leading to at least 1.3 self-renewing cells. We highlight an example of a blood-derived single cell phylogeny with major subclones separated by just two cell divisions. Broadly, our approach allows inference of tumor initiation and growth parameters based on events preceding the most recent common ancestor of the initiating clone as opposed to characteristics of fully grown tumors.

cancer biology↗

bioRxiv2025

Spatially organized lymphocytic microenvironments in high grade primary prostate tumors

The spatial arrangement of immune cells in the tumor microenvironment (TME) varies widely, from dispersed to clustered and tumor excluded to infiltrating. Multiplexed spatial profiling is an effective means of characterizing tumor-infiltrating lymphocytes (TILs) and immune complexes such as tertiary lymphoid structures (TLS) in the TME. However, few approaches have been described for objectively parametrizing patterns of immune organization and assessing their association with biological or clinical variables. This makes it difficult to evaluate whether a set of tumors is relatively immunologically cold or hot. Here we describe an intuitive set of statistical tools (available in the R package, tlsR) for characterizing lymphocyte patterns in the TME of solid cancers. We apply tlsR to primary prostate cancer (PCa), which is often described as immunologically cold. Using a cohort of 29 radical prostatectomy specimens stratified into low Gleason-grade (LGG; n=15) and high Gleason-grades (HGG; n =14) we show that HGG PCa is significantly more infiltrated than LGG PCa with lymphocytes organized into B cell or T cell enriched immune clusters (BICs and TICs). A subset of these ICs have the B and T cell zonation and follicular dendritic cells characteristic of a bona fide TLS. HGGs are also enriched with ICs containing precursor exhausted T cells (Tpex) and proliferating B cells and their tumor compartments harbor granzyme-B+ cytotoxic T cells in contact with cancer cells. Thus, far from being cold, a subset of HGG PCa has features associated with active immune surveillance, a finding with implications for emerging PCa immunotherapies.

cancer biology↗

bioRxiv2025

Disruption of Microhomology-mediated End-joining in Ewing Sarcoma

Ewing sarcoma (EwS) is a group of bone and soft tissue cancers in children and young adults. Since EwS cells have pronounced sensitivity to radiation and chemotherapy-induced DNA damage, the role of the oncoprotein, EWS-FLI1, in DNA repair is likely. Here, we demonstrate that EWS-FLI1 causes a defect in microhomology-mediated end-joining (MMEJ) repair. EWSR1 is a splicing factor that promotes the faithful splicing of the POLQ pre-mRNA, required for the expression of POL{Theta}, a critical protein in the MMEJ pathway. Expression of EWS-FLI1, or loss of EWSR1, causes exon 25 skipping of the POLQ transcript, decreased POL{Theta} expression, impaired MMEJ, and cellular sensitivity to inhibitors of the Fanconi Anemia (FA), NHEJ, or HR pathways, through the mechanism of synthetic lethality. Knockdown of EWS-FLI1 expression restores POL0 mitotic foci and increases MMEJ activity. Inhibitors of the FA, NHEJ, or HR therefore may provide a targeted therapy for patients with EwS. Highlights- Ewing sarcoma tumors have a deficiency in POL{theta} expression and a corresponding loss of MMEJ activity - EWSR1 is a splicing factor that interacts with other splicing factors such as FUBP1 and KHSRP/FUBP2 to accurately splice the POLQ mRNA. - The EWS-FLI1 fusion oncoprotein, or loss of EWSR1, causes a splicing defect, leading to exon25 skipping of the POLQ pre-mRNA and loss of POL{theta} expression - The MMEJ deficiency of EwS cells results in cellular sensitivity to inhibitors of Fanconi Anemia, Homologous Recombination or Non-Homologous End-Joining - Exon 25 skipping of POLQ mRNA is a predictive biomarker for HR inhibitors in human cancers

cancer biology↗

bioRxiv2024

Cell states and neighborhoods in distinct clinical stages of primary and metastatic esophageal adenocarcinoma

Esophageal adenocarcinoma (EAC) is a highly lethal cancer of the upper gastrointestinal tract with rising incidence in western populations. To decipher EAC disease progression and therapeutic response, we performed multiomic analyses of a cohort of primary and metastatic EAC tumors, incorporating single-nuclei transcriptomic and chromatin accessibility sequencing, along with spatial profiling. We identified tumor microenvironmental features previously described to associate with therapy response. We identified five malignant cell programs, including undifferentiated, intermediate, differentiated, epithelial-to-mesenchymal transition, and cycling programs, which were associated with differential epigenetic plasticity and clinical outcomes, and for which we inferred candidate transcription factor regulons. Furthermore, we revealed diverse spatial localizations of malignant cells expressing their associated transcriptional programs and predicted their significant interactions with microenvironmental cell types. We validated our findings in three external single-cell RNA-seq and three bulk RNA-seq studies. Altogether, our findings advance the understanding of EAC heterogeneity, disease progression, and therapeutic response.

cancer biology↗

bioRxiv2024

Rare germline structural variants increase risk for pediatric solid tumors

AO_SCPLOWBSTRACTC_SCPLOWPediatric solid tumors are rare malignancies that represent a leading cause of death by disease among children in developed countries. The early age-of-onset of these tumors suggests that germline genetic factors are involved, yet conventional germline testing for short coding variants in established predisposition genes only identifies pathogenic events in 10-15% of patients. Here, we examined the role of germline structural variants (SVs)--an underexplored form of germline variation--in pediatric extracranial solid tumors using germline genome sequencing of 1,766 affected children, their 943 unaffected relatives, and 6,665 adult controls. We discovered a sex-biased association between very large (>1 megabase) germline chromosomal abnormalities and a four-fold increased risk of solid tumors in male children. The overall impact of germline SVs was greatest in neuroblastoma, where we revealed burdens of ultra-rare SVs that cause loss-of-function of highly expressed, mutationally intolerant, neurodevelopmental genes, as well as noncoding SVs predicted to disrupt three-dimensional chromatin domains in neural crest-derived tissues. Collectively, our results implicate rare germline SVs as a predisposing factor to pediatric solid tumors that may guide future studies and clinical practice.

genomics↗

bioRxiv2023

Somatic structural variants driving distinct modes of oncogenesis in melanoma

The diversity of structural variants (SVs) in melanoma and how they impact oncogenesis are incompletely known. We performed harmonized analysis of SVs across melanoma histological and genomic subtypes, and we identified distinct global properties between subtypes. These included the frequency and size of SVs and SV classes, their relation to chromothripsis events, and the role of topologically associated domain (TAD) boundary altering SVs on cancer-related genes. Following our prior identification of double-stranded break repair deficiency in a subset of triple wild-type cutaneous melanoma, we identified MRE11 and NBN loss-of-function SVs in melanomas with this mutational signature. Experimental knockouts of MRE11 and NBN, followed by olaparib cell viability assays in melanoma cells, indicated that dysregulation of each of these genes may cause sensitivity to PARPi in cutaneous melanomas. Broadly, harmonized analysis of melanoma SVs revealed distinct global genomic properties and molecular drivers, which may have biological and therapeutic impact. Statement of SignificanceThe diversity of SVs in melanoma, and how they directly or indirectly impact oncogenesis, are incompletely known. Here we present analysis of melanoma SVs that reveal distinct global genomic properties and molecular drivers, some of which point to opportunities for further biological and therapeutic investigation.

cancer biology↗

bioRxiv2023

Analysis of evolutionary dynamics and clonal architecture in prostate cancer

The extent to which clinical and genomic characteristics associate with prostate cancer clonal architecture, tumor evolution, and therapeutic response remains unclear. Here, we reconstructed the clonal architecture and evolutionary trajectories of 845 prostate cancer tumors with harmonized clinical and molecular data. We observed that tumors from patients who self-reported as Black had more linear and monoclonal architectures, despite these men having higher rates of biochemical recurrence. This finding contrasts with prior observations relating polyclonal architecture to adverse clinical outcomes. Additionally, we utilized a novel approach to mutational signature analysis that leverages clonal architecture to uncover additional cases of homologous recombination and mismatch repair deficiency in primary and metastatic tumors and link the origin of mutational signatures to specific subclones. Broadly, prostate cancer clonal architecture analysis reveals novel biological insights that may be immediately clinically actionable and provide multiple opportunities for subsequent investigation. Statement of significanceTumors from patients who self-reported as Black demonstrate linear and monoclonal evolutionary trajectories yet experience higher rates of biochemical recurrence. In addition, analysis of clonal and subclonal mutational signatures identifies additional tumors with potentially actionable alterations such as deficiencies in mismatch repair and homologous recombination.

cancer biology↗

bioRxiv2022

Mutational footprint of platinum chemotherapy in a secondary thyroid cancer

Although papillary thyroid carcinoma (PTC) is the most frequent endocrine tumor with a generally excellent prognosis, a patient developed a clinically aggressive PTC eleven years after receiving platinum chemotherapy for ovarian endometrioid adenocarcinoma. Germline and somatic analyses of multi-temporal and multi-regional molecular profiles indicated that ovarian and thyroid tumors did not share common genetic alterations. PTC tumors had driver events associated with aggressive PTC behavior, an RBPMS-NTRK3 fusion and a TERT promoter mutation. Spatial and temporal genomic heterogeneity analysis indicated a close link between anatomical locations and molecular patterns of PTC. Mutational signature analyses demonstrated a molecular footprint of platinum exposure, and that aggressive molecular drivers of PTC were linked to prior platinum-associated mutagenesis. This case provides a direct association between platinum chemotherapy exposure and secondary solid tumor evolution, in specific aggressive thyroid carcinoma, and suggests that uniform clinical assessments for secondary PTC after platinum chemotherapy may warrant further evaluation.

genomics↗

bioRxiv2022

Patterns of Structural Variation Define Prostate Cancer Across Disease States

The complex genomic landscape of prostate cancer evolves across disease states under therapeutic pressure directed toward inhibiting androgen receptor (AR) signaling. While significantly altered genes in prostate cancer have been extensively defined, there have been fewer systematic analyses of how structural variation shapes the genomic landscape of this disease across disease states. We uniformly characterized structural alterations across 278 localized and 143 metastatic prostate cancers profiled by whole genome and transcriptome sequencing. We observed distinct significantly recurrent breakpoints in localized and metastatic castration-resistant prostate cancers (mCRPC), with pervasive alterations in noncoding regions flanking the AR, MYC, FOXA1, and LSAMP genes enriched in mCRPC and TMPRSS2-ERG rearrangements enriched in localized prostate cancer. We defined nine subclasses of mCRPC based on signatures of structural variation, each associated with distinct genetic features and clinical outcomes. Our results comprehensively define patterns of structural variation in prostate cancer and identify clinically actionable subgroups based on whole genome profiling.

genomics↗

bioRxiv2021

Dissecting tumor cell programs through group biology estimation in clinical single-cell transcriptomics

Given the growing number of clinically integrated cancer single-cell transcriptomic studies, robust differential enrichment methods for gene signatures to dissect tumor cellular states for discovery and translation are critical. Current analysis strategies neither adequately represent the hierarchical structure of clinical single-cell transcriptomic datasets nor account for the variability in the number of recovered cells per sample, leading to results potentially confounded by sample-driven biology with high false positives instead of accurately representing true differential enrichment of group-level biology (e.g., treatment responders vs. non-responders). This problem is especially prominent for single-cell analyses of the tumor compartment, because high intra-patient similarity (as opposed to inter-patient similarity) results in stricter hierarchical structured data that confounds enrichment analysis. Furthermore, to identify signatures which are truly representative of the entire group, there is a need to quantify the robustness of otherwise statistically significant signatures to sample exclusion. Here, we present a new nonparametric statistical method, BEANIE, to account for these issues, and demonstrate its utility in two cancer cohorts stratified by clinical groups to reduce biological hypotheses and guide translational investigations. Using BEANIE, we show how the consideration of sample-specific versus group biology greatly decreases the false positive rate and guides identification of robust signatures that can also be corroborated across different cell type compartments.

bioinformatics↗

bioRxiv2021

CREB5 reprograms nuclear interactions to promote resistance to androgen receptor targeting therapies

Metastatic castration resistant prostate cancers (mCRPC) are treated with therapies that antagonize the androgen receptor (AR). Nearly all patients develop resistance to AR-targeted therapies (ART). Our previous work identified CREB5 as an upregulated target gene in human mCRPC that promoted resistance to all clinically-approved ART. The mechanisms by which CREB5 promotes progression of mCRPC or other cancers remains elusive. Integrating ChIP-seq and rapid immunoprecipitation and mass spectroscopy of endogenous proteins (RIME), we report that cells overexpressing CREB5 demonstrate extensive reprogramming of nuclear protein-protein interactions in response to the ART agent enzalutamide. Specifically, CREB5 physically interacts with AR, the pioneering actor FOXA1, and other known co-factors of AR and FOXA1 at transcription regulatory elements recently found to be active in mCRPC patients. We identified a subset of CREB5/FOXA1 co-interacting nuclear factors that have critical functions for AR transcription (GRHL2, HOXB13) while others (TBX3, NFIC) regulated cell viability and ART resistance and were amplified or overexpressed in mCRPC. Upon examining the nuclear protein interactions and the impact of CREB5 expression on the mCRPC patient transcriptome, we found CREB5 was associated with TGF{beta} and Wnt signaling and epithelial to mesenchymal transitions, implicating these pathways in ART resistance. Overall, these observations define the molecular interactions among CREB5, FOXA1, and pathways that promote ART resistance.

cancer biology↗