Research Radar — 2026-08-31
Methods & AI
Computational
nf_xpatial: A Reproducible Framework for Standardized Preprocessing and Clustering of Xenium Data
bioRxiv (Bioinformatics) Published 2026-08-29 Preprint DOI: 10.64898/2026.08.25.747147
spatial omics Xenium Nextflow clustering
Summary: nf_xpatial is a Nextflow pipeline for downstream Xenium analysis, covering quality control, filtering, normalization, multi-sample integration, and expression-driven or spatially informed clustering across parameter sweeps.
Why it matters: A standardized, reproducible Xenium workflow addresses a practical bottleneck in spatial-omics analysis and makes clustering choices comparable across parameter settings.
Why for Yiru: This is directly relevant to spatial-omics preprocessing, integration, and downstream cellular-state analysis.
Visual LLM-guided consensus spatial domain detection with L-STAR
bioRxiv (Bioinformatics) Published 2026-08-29 Preprint DOI: 10.64898/2026.08.25.747158
spatial omics domain detection visual LLMs robustness
Summary: L-STAR uses visual large-language-model guidance to rank and integrate spatial domain detection methods, reporting more robust performance than individual methods across diverse datasets.
Why it matters: Consensus integration may improve robustness when spatial-domain detectors behave differently across datasets, although this remains a preprint claim to evaluate carefully.
Why for Yiru: The method is a direct fit for spatial-domain analysis and for testing whether visual-LLM guidance transfers across tissue contexts.
OmicsFM brings proteomics into the foundation model era
bioRxiv (Bioinformatics) Published 2026-08-28 Preprint DOI: 10.64898/2026.08.25.747021
proteomics foundation models multimodal biology transfer learning
Summary: OmicsFM pretrains a modality-agnostic transformer on 48,837 quality-filtered proteomics profiles from 1,397 PRIDE projects. Its representations support held-out-project analysis and transfer to cell-type classification, gene-essentiality, and perturbation-response prediction.
Why it matters: The scale of the proteomics corpus and held-out-project evaluations make a useful case for foundation-model representations beyond transcriptomics.
Why for Yiru: Its cross-modal framing is relevant to building transferable representations from proteomics and transcriptomics.
FOCUS-3D: Robust, generalizable volumetric cell segmentation for three-dimensional fluorescence microscopy
bioRxiv (Bioinformatics) Published 2026-08-28 Preprint DOI: 10.64898/2026.08.25.746907
3D microscopy cell segmentation computer vision spatial biology
Summary: FOCUS-3D combines volumetric representation learning, multiscale features, and query-based mask prediction for three-dimensional cell segmentation across species, tissues, reporters, and imaging modalities, with a zebrafish notochord application.
Why it matters: Cross-species and cross-modality segmentation generalization is important for turning tissue-scale microscopy into quantitative biology.
Why for Yiru: The framework is relevant to volumetric imaging and to connecting cellular morphology with spatial and developmental programs.
Closing the fusion-detection gap in single-cell RNA-seq with a scalable, probe-based workflow
bioRxiv (Bioinformatics) Published 2026-08-29 Preprint DOI: 10.64898/2026.08.26.747171
single-cell omics spatial transcriptomics gene fusions oncology
Summary: A probe-based workflow adds expressed oncogenic fusion detection to standard 10x Genomics Flex and Visium assays, recovering fusion counts alongside whole-transcriptome profiles in MCF7 cells and pediatric B-ALL cohorts.
Why it matters: The probe design bridges structural-variant detection with single-cell and spatial assays, making fusion-positive populations measurable in their transcriptomic context.
Why for Yiru: It is a practical example of adding oncology-relevant molecular specificity to spatial and single-cell workflows.
Biomedical discoveries
Biomedicine
Functional spatial transcriptomics uncover LMO7 as a fusion-regulated and clinically relevant driver of metastasis in Ewing sarcoma
bioRxiv (Cancer Biology) Published 2026-08-28 Preprint DOI: 10.64898/2026.08.27.747513
spatial oncology Ewing sarcoma metastasis functional genomics
Summary: Functional spatial transcriptomics of Ewing sarcoma identifies an invasive-front state with lower FET::ETS activity and higher LMO7 expression. Clinical, proteomic, and perturbation analyses implicate LMO7 in EMT, cytoskeletal remodeling, and metastasis.
Why it matters: The study connects a spatially localized tumor state to clinical outcome and experimentally testable metastasis biology.
Why for Yiru: It closely matches interests in spatial oncology, multimodal integration, and perturbation-linked disease-state discovery.
Disrupting Myeloid Persistence and Replenishment Enables Sustained Control of Esophageal Squamous Cell Carcinoma
bioRxiv (Cancer Biology) Published 2026-08-28 Preprint DOI: 10.64898/2026.08.27.747541
immuno-oncology myeloid cells single-cell omics therapy resistance
Summary: In esophageal squamous cell carcinoma, CSF1R inhibition removes established tumor-associated macrophages but permits complementary monocytic and granulocytic replenishment. Combined treatment with low-dose decitabine achieves sustained control across organoid xenograft, orthotopic, and immunocompetent models.
Why it matters: The treatment-resolved myeloid architecture explains a concrete route to therapeutic escape and motivates state-aware combination strategies.
Why for Yiru: Its single-cell view of immune niches and therapy response is directly useful for computational immuno-oncology.
An Open Benchmark for Systems Vaccinology: Insights from the CMI-PB Challenges
bioRxiv (Immunology) Published 2026-08-28 Preprint DOI: 10.64898/2026.08.25.746820
computational immunology systems vaccinology benchmarking multi-omics
Summary: The CMI-PB challenges benchmark 107 models predicting pertussis booster responses from pre-booster multimodal data, testing generalization on an unseen cohort and identifying immune setpoints, preprocessing, imputation, and multi-omics integration as important factors.
Why it matters: Prospective evaluation on an unseen cohort is a valuable test of whether predictive-immunology models generalize beyond training data.
Why for Yiru: The benchmark offers a transferable template for multimodal immune-response modeling and rigorous preprocessing.
Extracellular protein catabolism drives regulated nitrogen handling and ammonia buffering in acute myeloid leukemia
bioRxiv (Cancer Biology) Published 2026-08-28 Preprint DOI: 10.64898/2026.08.27.747222
cancer metabolism acute myeloid leukemia metabolomics therapeutic targets
Summary: AML cells use lysosomal albumin degradation as an amino-acid source, generating ammonia that is buffered by glutamate-ammonia ligase (GS/GLUL). Metabolomics, isotope tracing, and perturbation experiments identify GS capacity as a vulnerability of proteocatabolic growth.
Why it matters: The work links a measurable metabolic adaptation to a defined nitrogen-buffering dependency and tests that dependency in disease models.
Why for Yiru: It is relevant to cancer metabolism and to integrating metabolomics with perturbation and therapeutic hypotheses.
Single-cell mapping of the fallopian tube reveals a genomically unstable secretory cell state enriched in carriers of germline BRCA1/2 mutations
bioRxiv (Cancer Biology) Published 2026-08-27 Preprint DOI: 10.64898/2026.08.26.746998
single-cell omics fallopian tube BRCA1/2 cancer interception
Summary: Single-cell, multi-regional profiling of fallopian tubes from 34 women, including 15 germline BRCA1/2 carriers, identifies a TP53-high secretory-cell state with replication-stress features enriched in carriers and supported by gammaH2AX and 53BP1 protein validation.
Why it matters: A carrier-enriched, genomically unstable secretory state offers a candidate early-interception signal supported by orthogonal protein measurements.
Why for Yiru: The study is a strong example of single-cell disease-state discovery with direct relevance to inherited cancer risk.
Phase-resolved transcriptomic bottlenecks in peptide cancer vaccine response
bioRxiv (Immunology) Published 2026-08-26 Preprint DOI: 10.64898/2026.08.21.746349
computational immunology cancer vaccines transcriptomics biomarkers
Summary: Across three public peptide-vaccine cohorts, phase-linked transcriptomic modules associate baseline immune readiness, erythroid/inflammatory signals, dendritic-cell product state, and early priming with response-related endpoints. The retrospective analyses do not establish causality, clinical utility, or durable tumor control.
Why it matters: The cautious phase-linked analysis separates response-associated layers from claims of causality, a useful discipline for retrospective immunology.
Why for Yiru: It provides a computational framework for thinking about immune readiness and vaccine-response heterogeneity.
Glutamatergic Neuron-Meningioma Synapse Interaction Promotes Brain-Invasive Tumor Growth
bioRxiv (Cancer Biology) Published 2026-08-27 Preprint DOI: 10.64898/2026.08.26.747240
neuro-oncology meningioma tumor–neuron interactions electrophysiology
Summary: Electron microscopy, single-cell transcriptomics, electrophysiology, and intracranial xenografts provide evidence for functional neuron-meningioma communication. Glutamatergic signaling and AMPA/NMDA receptor blockade affect proliferation, particularly in brain-invasive tumor biology.
Why it matters: The multimodal evidence supports a mechanistic link between neuronal signaling and aggressive meningioma growth, while leaving the preprint findings to further validation.
Why for Yiru: It is relevant to tumor–neuron interactions and to integrating spatial, electrophysiological, and single-cell evidence.
A PTBP1-CDC42 splicing axis regulates leukemia growth and venetoclax sensitivity in acute myeloid leukemia
bioRxiv (Cancer Biology) Published 2026-08-26 Preprint DOI: 10.64898/2026.08.25.745954
acute myeloid leukemia RNA splicing drug response single-cell biology
Summary: Integrative transcriptomic and iCLIP analyses identify PTBP1 control of a CDC42 splice switch in AML. CDC42 inhibition selectively affects leukemia cells and enhances venetoclax efficacy in the reported models.
Why it matters: The PTBP1–CDC42 mechanism links RNA regulation to leukemia fitness and suggests a combination strategy with venetoclax.
Why for Yiru: It combines transcriptomic regulation, perturbation, and therapeutic response in a computationally tractable oncology setting.