Research Week Core Technology Talks Abstracts

Screening Therapies Against Living Patient Tumor Tissues

Screening Live Cancer Explants (SLICE) Core

Dr. Andrew Satterlee

We have developed a platform that can directly and rapidly screen therapeutics against diverse patient tumor tissues to measure patient-specific drug sensitivities: Screening Live Cancer Explants (SLiCE) Technology. SLiCE engrafts heterogeneous, passage-zero patient tumor tissues atop living tissue substrates and recapitulates clinically relevant tumor phenotypes that include growth, invasion, and response to treatment. Through our actively accruing clinical feasibility studies, including NCT05978557, we have successfully captured 47-of-47 high- and low-grade, adult and pediatric patient tumor tissues with high retention of genomic and cellular heterogeneity. Our AI-driven multi-parametric readouts of therapeutic activity and off-target toxicity for diverse classes of drugs ‚Äì including effector cell therapies such as CAR T cells ‚Äì provide functional readouts that are quantified and analyzed in as few as five days after surgery. Our standard protocol also includes a cryopreservation step and can freeze tumors in multiple representative aliquots, enabling patient tumor shipment from other sites to our central location for on-demand testing. Our Screening Live Cancer Explants Core Facility at UNC now enables other groups to utilize the SLiCE Human-Based Lab Model to accelerate translation of their novel therapeutics.

Olink High-Throughput Biomarker Analysis at UNC- a Long COVID case study

Respiratory TRACTS Core

Mandy Bush

Olink is a high-throughput biomarker analysis platform that utilizes ground-breaking Proximity Extension Assay (PEA) technology to enable scalable discovery, development, and validation of protein signatures. This innovative platform allows researchers to investigate large numbers of protein expression profiles in very small sample volumes, enabling large-scale analysis on previously challenging samples. For this case study, we used serum samples from Long COVID patients that were recruited through the UNC COVID Recovery Clinic to investigate inflammatory biomarkers associated with Long COVID. Samples were derived from Tasso devices, which are notorious for low sample volumes and variable sample quality.

AI Meets Drug Discovery: Harnessing MolGPT for Designing Potent Protein Binders

R.L. Juliano Structural Bioinformatics Core

Dr. Venkata Reddy Chirasani

Advances in generative deep learning are transforming early-stage drug discovery by enabling targeted exploration of chemical space. In this work, we apply MolGPT, a scaffold- and property-conditioned generative transformer, to design novel small molecules with high predicted affinity for specific protein targets. The model was trained on diverse chemical datasets and fine-tuned on a curated set of experimental binder/non-binder SMILES to capture critical structure–activity relationships. By integrating scaffold-based conditioning with binding score control, we generate synthetically accessible, drug-like molecules tailored to target-specific features. Our results highlight the power of AI-guided molecular generation to accelerate the discovery of novel chemical matter for challenging therapeutic targets, paving the way for experimental synthesis and validation.

Obtain Collision Cross-Section Values in Cyclic Ion Mobility Separations

Biomarker Mass Spectrometry Core

Dr. Zhenfa Zhang and Evan Ritter

Ion mobility spectrometry-mass spectrometry (IMS-MS) is useful in the field of omics- study, particularly with the development of high-resolution platforms; The collision cross section (CCS) is one of the only quantitative parameters in IMS-MS. The determination of the CCS other outside of drift tube IMS systems is not straightforward. In cyclic ion mobility-mass spectrometry (cIMS MS), often relaying on calibration as opposed directly calculation. A general method that uses average ion velocities for calculating CCS values in cIMS-MS-based separations is reproduced.

Discover the Future of Histopathology with HALO AI at the Pathology Services Core

Pathology Services Core

Nicholas Pankow, Edison Floyd, and Dr. Hannah Atkins

Are you interested in adding more machine learning (ML) and artificial intelligence (AI) technology to your tissue-based research? The UNC Pathology Services Core (PSC) recently acquired HALO AI software from Indica Labs through a grant from the NC Biotechnology Center. The PSC offers a range of services, including clinical archiving, histology, whole-slide imaging (WSI), and next-generation spatial technology. Halo AI integrates into this workflow by utilizing advanced supervised ground-truth training to extract quantitative data from sets of brightfield or fluorescent WSIs created from paraffin-embedded, formalin-fixed tissue. For example, HALO AI can identify and quantify the percentage of cells positive for specific immunohistochemical biomarkers in isolated tissues or regions, as well as analyze the spatial relationships among different cell populations. HALO AI features over 15 pre-trained modules designed to accelerate analysis and enhance accuracy. As with any histology-based method, quality control (QC) is essential throughout all stages of preparation for analysis. However, the AI and ML tools provided by HALO AI offer many ways to address QC challenges when slide quality is outside the investigator‚Äôs control, which is often the case with clinical archival studies. Overall, utilizing HALO AI and image analysis to collect quantitative data can significantly enhance the rigor and reproducibility of tissue-based research. Contact the PSC Facility Director, Gaby De la Cruz, and the PSC Facility Manager, Lauren Harris, to learn how AI and ML can advance your research!

Clinical proteomics for the masses

Michael Hooker Metabolomics and Proteomics Core

Dr. Laura Herring

Plasma is a rich source of biomarkers that can inform therapeutic decision-making, predict treatment response and identify mechanisms of resistance or therapeutic vulnerabilities in cancer. Recent advancements in mass spectrometry-based proteomics, including sample preparation, automation, instrumentation, and data analytics, have significantly enhanced plasma proteomic throughput, reproducibility, and depth. These developments have the potential to refine biomarker-driven strategies for precision oncology. We performed a comparison of four plasma sample preparation workflows (neat, acid depletion, extracellular vesicle enrichment, and nanoparticle enrichment) in a breast cancer patient cohort pre/post neoadjuvant chemotherapy (NACT) treatment. The enrichment strategies provided the greatest proteomic depth, while neat and acid depletion methods achieved the highest throughput. All methods demonstrated high reproducibility, highlighting their feasibility for clinical implementation.

Decoding Biological Complexity through Multi-Omics

High-Throughput Sequencing Facility

Dr. Piotr Mieczkowski

Advancing our understanding of complex biological systems requires access to high-quality data generated across multiple molecular layers. At UNC, the integration of campus-wide core facilities enables the generation of comprehensive multi-omic datasets, encompassing transcriptomics, proteomics, metabolomics, lipidomics, and DNA methylation. Each of these data types is produced using specialized technologies housed in dedicated facilities: high-throughput RNA-seq, Small RNA-seq and epigenetic mapping via Enzymatic Methylation sequencing from the High Throughput Sequencing Facility (HTSF); mass spectrometry-based proteomics from the Michael Hooker Proteomics, Center Mass Spectrometry Core (Chemistry) and Biomolecular NMR Core; targeted and untargeted metabolomic and lipidomic profiling from the Mass Spectrometry Core (Chemistry), Biomolecular NMR Core and Metabolomics Core. This talk will showcase how leveraging the collective capabilities of UNC‚Äôs research infrastructure empowers large-scale, systems-level studies in non-traditional model organisms. Using examples from ongoing research in Eptesicus fuscus (big brown bat), we will highlight how coordinated workflows across facilities support the generation of high-resolution molecular data from multiple tissues and physiological states. These efforts not only enable deeper biological insights but also provide a platform for collaborative research, student training, and future applications of AI-driven analytics. By highlighting the strengths of UNC’s core resources, this presentation underscores the critical role of institutional infrastructure in enabling cutting-edge, multi-omics research.

Getting the most RNA from your blood

Biospecimen Processing Facility

Christopher Nagy

The use of PAXGene RNA Blood tubes is the gold standard for isolating RNA from whole blood, however, cchieving a high-quality RNA result is highly dependent upon how the blood samples are treated in the PAXGene tubes prior to extraction. Factors such as lack of mixing immediately post blood collection, or prolonged exposure to heat play a significant role in the formation of clots within the PAXGene blood tube and result in loss of quality and/or abundance of RNA in the final extraction.

Setting up a genomic long-read sequence analysis pipeline

Clinical Genomic Analysis (GENYSIS) Core

Bradford Powell and Scott Melville

The UNC Clinical Genomic Analysis (GENYSIS) core facility has routinely provided exome and genome analysis of short-read sequence data. However, as many of the GENYSIS cases for analysis have had previous negative clinical testing by short-read technologies, we were asked by the UNC clinical genomic research community if we could provide a service to analyze long-read sequence data. As the UNC High-Throughput Sequencing Facility (HTSF) offers Oxford Nanopore Technology (ONT) long-read sequencing, we set up a bioinformatic analysis pipeline using the open-source EPI2ME platform. We optimized the pipeline to allow for analysis of single nucleotide variants (SNVs), copy number variants (CNVs), structural variants (SVs) including insertions, deletions, duplications, inversions and translocations, and certain repeat expansions. Future work includes analysis of the methylation data that accompanies the long-read data.

Efficient and Cost-Effective 16S rRNA Sequencing: Preserving Microbial Diversity with Less Reagents

UNC Microbiome Core

Nivedita Pareek

High-throughput next-generation sequencing (NGS) has transformed microbiome research by allowing detailed analysis of microbial communities. However, library preparation remains a major cost factor, accounting for up to 75% of total sequencing expenses, mainly due to reagent costs. A useful approach to cut costs involves reducing reagent volumes and increasing sample throughput. Previous studies have shown that downsizing reaction volumes can lower costs significantly without affecting data quality.

We evaluated a miniaturized 16S rRNA library preparation protocol using the Opentrons Flex robot to assess the feasibility of reducing reaction volumes while maintaining data integrity. Libraries were prepared using the Illumina protocol in two formats: half- reaction volumes prepared manually, and quarter-reaction volumes prepared robotically. Out of 96 stool samples processed for MiSeq PE 250 sequencing, 18 were selected for direct comparison. There was no significant difference in input or non-chimeric read counts between the two methods. Alpha diversity metrics (Faith‚Äôs PD and Shannon) showed no statistically significant differences between half and quarter reactions (p = 0.7646 and p = 0.6286, respectively), indicating similar microbial richness and evenness. Beta diversity analysis via PCoA revealed no distinct clustering between the two groups, suggesting the overall community structure remained consistent. Taxonomic profiles at the phylum and genus levels also showed high concordance across preparations. No taxa were identified, neither pairwise nor in aggregate, as significantly different in abundance. Our data confirms that providing high-quality, cost-effective microbiome sequencing services is achievable despite increasing reagent and sequencing costs, by decreasing reagent usage per sample without sacrificing data integrity.

Achieving High-Resolution Light-Sheet Imaging with the ASI ct-dSPIM

Neuroscience Microscopy Core

Dr. Michelle Itano and Tessa Ropp

We will present the impact and potential of the ct-dSPIM imaging system in the Neuroscience Microscopy Core. This imaging system, installed with the support of an instrumentation grant from the NIMH, enables imaging of large samples or at higher speeds with improved resolution in the z-dimension. Recent advances from 3D printed sample holders to customized image processing coding, have enabled the development of a full pipeline for either advanced user training for independent use of the system, or fee-for-service imaging with support from core staff.

Mixed Method Evaluation: Case studies from the ECHO program

Abacus Evaluation

Tara Carr, Shelly Maras, and Jade Hollars

Health and wellbeing programs and interventions, especially complex and national ones, require comprehensive, rigorous, and mixed methods evaluation to assess their effectiveness in achieving expected outcomes and creating impacts which can improve human health and wellbeing. Organizations and researchers often look to partner with external third-party evaluators for an unbiased appraisal to guide them in developing and implementing mixed methods evaluations which use multiple approaches and methodologies to evaluate program and intervention components individually and comprehensively. We will present a case study as an example of how Abacus designed a mixed-methods evaluation for the National Institutes of Health (NIH) — Environmental influences on Child H