Our Science
Our approach
T cells help control cancer through sensing DNA mutations by activating when neoantigens are presented in the major histocompatibility complex (MHC) via a protein called the T cell receptor (TCR). For some patients with cancer and other rare diseases, the only prospective treatment is an allogeneic hematopoietic cell transplant (allo-HCT), commonly called a bone marrow transplant (BMT), which introduces another person’s T cells into the body. Until now, diagnostic tests have been limited in the ability to predict safe and effective outcomes of this treatment, focusing solely on the MHC, which too often leads to cancer relapse or a devastating and life-threatening allo-immune condition called graft-versus-host disease (GVHD).
ImmunoScope is developing predictive diagnostic tests focused on TCR sequencing and anchored by the fundamental immunological process of T cell selection. Our lead tests, the Novel Fraction and the Tolerant Fraction are developed with pre-transplant biospecimens from BMT donor-recipient pairs and validated using matched clinical outcomes data to predict cancer relapse and GVHD, respectively. We are advancing the Novel and Tolerant Fraction tests alongside hematologists, aiming to deliver actionable, pre-transplant immunological insights, helping guide optimal treatment decisions and patient outcomes.
Biophysicochemical motifs in T cell receptor sequences as a potential biomarker for high-grade serous ovarian carcinoma.
March 5, 2020
Prior work showed that specific biophysicochemical motifs in TCR antigen-binding regions distinguish tumor-infiltrating lymphocyte repertoires from adjacent healthy tissue. These cancer-specific motifs classified breast and colorectal samples with high accuracy (94% and 93%, respectively), suggesting a path to TCR-based cancer detection. Here, we investigate whether analogous motifs exist for ovarian cancer, where better detection tools are urgently needed.
Advances in immune repertoire sequencing unlock rich insights into adaptive immunity but generate massive, complex datasets that demand specialized analysis. We built VDJServer to provide an end-to-end pipeline with integrated data management. It also delivers accessible high-performance computing, lowering technical barriers for researchers.
Deep repertoire sequencing now lets us profile lymphocyte clones in depth, enabling diagnostic and prognostic signals tied to immune-driven disease. Prior approaches relied on coarse repertoire summaries, overlooking information in millions of individual receptors. We introduce a method that handles extreme sequence diversity with standard machine learning and apply it to RRMS.
T-cell tolerant fraction as a predictor of immune-related adverse events.
August 13, 2023
Immune checkpoint inhibitors can trigger unpredictable autoimmune toxicities (irAEs), and T-cell profiling may help identify who’s at risk. We introduce the T-cell “tolerant fraction”—derived from productive vs. non-productive TCRβ sequences—as a new repertoire metric from pre-treatment blood. This tolerant fraction was associated with subsequent irAE development across multiple cancers and ICI regimens.
Reconstituting T cell receptor selection in-silico.
June 14, 2021
TCRs are generated without regard to antigen, and thymic selection removes cells that can’t engage MHC or bind self too strongly—but which sequences are kept vs. removed has been unclear. Using TCR sequencing and machine learning, we identify sequence features that predict selection outcomes and accurately distinguish pre-selection (developing) from post-selection (mature) TCRs. This approach lays groundwork for studying how selection shapes immunity and its links to autoimmune disease.
Cervical cancer screening relies on Pap and hrHPV tests that miss key risk distinctions, leading to both overtreatment and missed cancers. Using T-cell receptor deep sequencing and machine learning, we identified a CDR3β motif and built a classifier from index cervical cytology that predicted transition to lower risk with 95% accuracy (19/20, leave-one-out). Pending prospective validation, this approach could sharpen risk stratification—reducing surveillance and overtreatment in screened populations and improving screening delivery for under-screened groups.
Statistical classifiers for diagnosing disease from immune repertoires: a case study using multiple sclerosis.
September 7, 2017
Publications
We introduce the TRB tolerant fraction, a novel metric that quantifies how well donor TCR$\beta$ sequences align within a recipient’s specific tolerance space. By comparing donor sequences that overlap with the recipient's productive versus non-productive repertoires, this method effectively isolates signals of T-cell tolerance. Validated in both murine models and human allo-HCT cohorts, the tolerant fraction consistently identifies the risk of graft-versus-host disease (GvHD). Ultimately, this metric provides critical predictive value that extends beyond routine clinical factors.
VDJServer: A Cloud-Based Analysis Portal and Data Commons for Immune Repertoire Sequences and Rearrangements.
May 8, 2018
Dynamic kernel matching for non-conforming data: A case study of T cell receptor datasets.
March 7, 2023
Most classifiers expect spreadsheet-style data, but many real datasets don’t fit that mold. We introduce dynamic kernel matching (DKM), which adapts established classifiers to handle non-conforming data. Applied to T-cell receptor sequences and repertoires, DKM detects diagnostic signatures for antigen specificity and CMV serostatus.
T Cell Receptor Repertoires Acquired via Routine Pap Testing May Help Refine Cervical Cancer and Precancer Risk Estimates.
April 2, 2021
T cell tolerant fraction as a predictor of graft-vs-host disease following allogeneic hematopoietic cell transplantation
November 10, 2025