Single-cell RNA-seq (scRNA-seq) has emerged as a powerful technique for characterizing cellular heterogeneity, but it is currently impractical on large sample cohorts and cannot be applied to fixed specimens collected as part of routine clinical care. We previously developed an approach for digital cytometry, called CIBERSORT, that enables estimation of cell type abundances from bulk tissue transcriptomes. We now introduce CIBERSORTx, a machine learning method that extends this framework to infer cell-type-specific gene expression profiles without physical cell isolation. By minimizing platform-specific variation, CIBERSORTx also allows the use of scRNA-seq data for large-scale tissue dissection. We evaluated the utility of CIBERSORTx in multiple tumor types, including melanoma, where single-cell reference profiles were used to dissect bulk clinical specimens, revealing cell type-specific phenotypic states linked to distinct driver mutations and response to immune checkpoint blockade. We anticipate that digital cytometry will augment single-cell profiling efforts, enabling cost-effective, high-throughput tissue characterization without the need for antibodies, disaggregation, or viable cells.

Supplemental Digital Content is available in the text.

Cancer somatic mutations can generate neoantigens that distinguish malignant from normal cells 1–7. However, the personalized identification and validation of neoantigens remains a major challenge. Here we discover neoantigens in human mantle cell lymphomas using an integrated genomic and proteomic strategy that interrogates tumor antigen peptides presented by major histocompatibility complex (MHC) class I and class II molecules. We applied this approach to systematically characterize MHC ligands from 17 patients. Remarkably, all discovered neoantigenic peptides were exclusively derived from the lymphoma immunoglobulin (Ig) heavy or light chain variable regions. Although we identified MHC presentation of private polymorphic germline alleles, no mutated peptides were recovered from non-Ig somatically mutated genes. Somatic mutations within the immunoglobulin variable region were almost exclusively presented by MHC-II. We isolated circulating CD4 T-cells specific for immunoglobulin-derived neoantigens and found these cells could mediate killing of autologous lymphoma cells. These results demonstrate that an integrative approach combining MHC isolation, peptide identification and exome sequencing is an effective platform to uncover tumor neoantigens. Application of this strategy to human lymphoma implicates immunoglobulin neoantigens as targets for lymphoma immunotherapy.

Although rapid advances in machine learning have made it increasingly applicable to expert decision-making, the delivery of accurate algorithmic predictions alone is insufficient for effective human-AI collaboration. In this work, we investigate the key types of information medical experts desire when they are first introduced to a diagnostic AI assistant. In a qualitative lab study, we interviewed 21 pathologists before, during, and after being presented deep neural network (DNN) predictions for prostate cancer diagnosis, to learn the types of information that they desired about the AI assistant. Our findings reveal that, far beyond understanding the local, case-specific reasoning behind any model decision, clinicians desired upfront information about basic, global properties of the model, such as its known strengths and limitations, its subjective point-of-view, and its overall design objective-what it's designed to be optimized for. Participants compared these information needs to the collaborative mental models they develop of their medical colleagues when seeking a second opinion: the medical perspectives and standards that those colleagues embody, and the compatibility of those perspectives with their own diagnostic patterns. These findings broaden and enrich discussions surrounding AI transparency for collaborative decision-making, providing a richer understanding of what experts find important in their introduction to AI assistants before integrating them into routine practice. CCS Concepts: • Human-centered computing → Human computer interaction (HCI).

Background: Deep learning has the potential to augment the use of chest radiography in clinical radiology, but challenges include poor generalizability, spectrum bias, and difficulty comparing across studies.Purpose: To develop and evaluate deep learning models for chest radiograph interpretation by using radiologist-adjudicated reference standards. Materials and Methods:Deep learning models were developed to detect four findings (pneumothorax, opacity, nodule or mass, and fracture) on frontal chest radiographs. This retrospective study used two data sets. Data set 1 (DS1) consisted of 759 611 images from a multicity hospital network and ChestX-ray14 is a publicly available data set with 112 120 images. Natural language processing and expert review of a subset of images provided labels for 657 954 training images. Test sets consisted of 1818 and 1962 images from DS1 and ChestX-ray14, respectively. Reference standards were defined by radiologist-adjudicated image review. Performance was evaluated by area under the receiver operating characteristic curve analysis, sensitivity, specificity, and positive predictive value. Four radiologists reviewed test set images for performance comparison. Inverse probability weighting was applied to DS1 to account for positive radiograph enrichment and estimate population-level performance.