Biology subjects

Kleinstein, S.

Publications and source records attributed to Kleinstein, S..

18 recordsLinked to original sources

bioRxiv2026

Cell-level integration of gene expression and B cell receptor data captures complementary B cell information

Single-cell B cell analyses commonly define relationships among cells using gene expression (GEX), although paired B cell receptor (BCR) sequences provide complementary information about receptor features, differentiation, and clonal relationships. We evaluated multiple approaches for integrating paired GEX and BCR data into cell-level representations and compared their ability to capture complementary aspects of B cell biology. We compared BCR representations based on language-model embeddings and engineered sequence features across multiple integration strategies. Spaces dominated by BCR language-model embeddings captured broad sequence and clonal information but could distort cell type relationships, as assessed using independent cell surface protein measurements. In contrast, concatenating GEX with selected BCR features (isotype, somatic hypermutation frequency, and CDRH3 length) incorporated receptor information while retaining transcriptional and surface-protein-associated structure. The resulting space reduced discordance between GEX-based cell type annotations and BCR characteristics and revealed B-cell substructure supported by both transcriptional and receptor-derived features. Thus, integrating GEX and BCR data captures complementary biological information and provides a richer foundation for single-cell B cell analysis.

bioinformatics↗

bioRxiv2026

Assessing tensor decomposition quality of immune profiling data from a dictionary learning perspective

Systems-level immune profiling data arising from longitudinal studies of vaccination or infection has an inherent multi-index array structure. While tensor decomposition of such datasets has gained popularity, choosing a rank and trial for a decomposition is not straightforward. We show that taking into account the experimental data model can inspire the development of new metrics to assess the quality of a Non-negative CANDECOMP/PARAFAC (NCPD) decomposition, and can thus be used to choose a rank and trial for the decomposition. Moreover, we show how framing the results via a dictionary learning framework can better enable interpretation of the components of the decomposition.

bioinformatics↗

bioRxiv2026

Robust semi-supervised scRNA-seq integration from virtual adversarial learning

Single-cell RNA sequencing integration methods that rely solely on transcriptomic data often struggle to preserve fine-grained distinctions between closely related cell subtypes. As a result, cell populations that are separable in the raw data may become over-mixed after integration, reducing biological resolution and interpretability. Incorporating marker gene information can potentially address these issues; however, the variability and complexity of available marker sets limit their effective application. To address this, we introduce scCRAFT+, a semi-supervised integration model that innovatively incorporates marker gene information through Virtual Adversarial Training (VAT). By jointly optimizing marker-derived supervision and transcriptome-wide representations, VAT enforces local prediction smoothness among transcriptionally similar cells, improving robustness to noisy marker annotations while enhancing both integration quality and cell type auto-annotation. This targeted approach significantly enhances annotation accuracy and robustness, particularly when faced with incomplete or incorrect marker gene sets. Benchmarking shows that scCRAFT+ achieves consistently stronger performance than current unsupervised and supervised integration approaches, resulting in improved integration quality and biologically meaningful sub-cell type auto-annotations.

bioinformatics↗

bioRxiv2025

AMULETY: A Python package to embed adaptive immune receptor sequences

Large language models have been developed to capture relevant features of adaptive immune receptors, each with unique potential applications. However, the diversity in available models presents challenges in accessibility and usability for downstream applications. Here we present AMULETY (Adaptive imMUne receptor Language model Embedding Tool), a Python-based software package to generate language model embeddings for adaptive immune receptor sequences, enabling users to leverage the strengths of different models without the need for complex configuration. AMULETY offers functions for embedding adaptive immune receptor amino acid sequences using pre-trained protein or antibody language models for paired heavy and light chain or single chain sequences. We showcase the variability on the embedding space for several embeddings on a dataset of antibody binders to several SARS-CoV-2 epitopes and showed that different models may be effective at capturing different aspects of the distinctions between epitope groups. AMULETY is available under GPL v3 license from https://github.com/immcantation/amulety or via pip from the Python Package Index (PyPI) from https://pypi.org/project/amulety/. O_FIG O_LINKSMALLFIG WIDTH=200 HEIGHT=164 SRC="FIGDIR/small/644583v1_ufig1.gif" ALT="Figure 1"> View larger version (33K): org.highwire.dtl.DTLVardef@1c0bafaorg.highwire.dtl.DTLVardef@337816org.highwire.dtl.DTLVardef@111136forg.highwire.dtl.DTLVardef@1300dc_HPS_FORMAT_FIGEXP M_FIG C_FIG

bioinformatics↗

bioRxiv2024

Partially characterized topology guides reliable anchor-free scRNA integration

Single-cell RNA sequencing (scRNA-seq) is an important technique for obtaining biological insights at cellular resolution, with scRNA-seq batch integration a key step before downstream statistical analysis. Despite the plethora of methods proposed, achieving reliable batch correction while preserving the heterogeneity of biological signals that define cell type continues to pose a challenge, with existing methods performance varying significantly across different scenarios and datasets. To address this, we propose scCRAFT, an autoencoder model designed to segregate cell-type-related biological signals from batch effects for reliable multi-batch scRNA-seq integration. scCRAFT comprises three key loss components: a reconstruction loss that targets observation reconstruction, a multi-domain adaptation loss aimed at eliminating batch effects, and an innovative dual-resolution triplet loss for preserving topology within each batch, which is introduced as an effective mechanism to counteract the over-correction effect of domain adaptation loss amid heterogeneous cell distributions across batches. We show that scCRAFT effectively manages unbalanced batches, rare cell types, and batch-specific cell phenotypes in simulations, and surpasses state-of-the-art methods in a diverse set of real datasets.

bioinformatics↗

bioRxiv2024

Dissecting human monoclonal antibody responses from mRNA- and protein-based XBB.1.5 COVID-19 monovalent vaccines

The emergence of highly contagious and immune-evasive severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) variants has required reformulation of coronavirus disease 2019 (COVID-19) vaccines to target those new variants specifically. While previous infections and booster vaccinations can enhance variant neutralization, it is unclear whether the monovalent version, administered using either mRNA or protein-based vaccine platforms, can elicit de novo B-cell responses specific for Omicron XBB.1.5 variants. Here, we dissected the genetic antibody repertoire of 603 individual plasmablasts derived from five individuals who received a monovalent XBB.1.5 vaccination either with mRNA (Moderna or Pfizer/BioNtech) or adjuvanted protein (Novavax). From these sequences, we expressed 100 human monoclonal antibodies and determined binding, affinity and protective potential against several SARS-CoV-2 variants, including JN.1. We then select two vaccine-induced XBB.1.5 mAbs, M2 and M39. M2 mAb was a de novo, antibody, i.e., specific for XBB.1.5 but not ancestral SARS-CoV-2. M39 bound and neutralized both XBB.1.5 and JN.1 strains. Our high-resolution cryo-electron microscopy (EM) structures of M2 and M39 in complex with the XBB.1.5 spike glycoprotein defined the epitopes engaged and revealed the molecular determinants for the mAbs specificity. These data show, at the molecular level, that monovalent, variant-specific vaccines can elicit functional antibodies, and shed light on potential functional and genetic differences of mAbs induced by vaccinations with different vaccine platforms. GRAPHICAL ABSTRACT O_FIG O_LINKSMALLFIG WIDTH=200 HEIGHT=43 SRC="FIGDIR/small/602781v1_ufig1.gif" ALT="Figure 1000"> View larger version (16K): org.highwire.dtl.DTLVardef@7c4708org.highwire.dtl.DTLVardef@11b66acorg.highwire.dtl.DTLVardef@1f1cec7org.highwire.dtl.DTLVardef@3e72fe_HPS_FORMAT_FIGEXP M_FIG C_FIG

immunology↗

bioRxiv2023

Antibody gene features associated with binding and functional activity in vaccine-derived human mAbs targeting malaria parasites

Adjuvants have been essential to malaria vaccine development, but their impact on the vaccine-induced antibody repertoire is poorly understood. Here, we used cDNA sequences from antigen-specific single memory B cells to express 132 recombinant human anti-Pfs230 monoclonal antibodies (mAbs). Alhydrogel(R)-induced mAbs demonstrated higher binding to Pfs230D1, although functional activity was similar between adjuvants. All Alhydrogel(R) mAbs using IGHV1-69 gene bound to recombinant Pfs230D1, but none blocked parasite transmission to mosquitoes; similarly, no AS01 mAb using IGHV1-69 blocked transmission. Functional mAbs from both Alhydrogel(R) and AS01 vaccines used IGHV3-21 and IGHV3-30 genes. Antibodies with the longest CDR3 sequences were associated with binding but not functional activity. This study assesses adjuvant effects on antibody clonotype diversity during malaria vaccination.

immunology↗

bioRxiv2023

Language model-based B cell receptor sequence embeddings can effectively encode receptor specificity

High throughput sequencing of B cell receptors (BCRs) is increasingly applied to study the immense diversity of antibodies. Learning biologically meaningful embeddings of BCR sequences is beneficial for predictive modeling and interpretability. Several embedding methods have been developed for BCRs, but no direct performance benchmarking exists. Moreover, the impact of the input sequence length and paired-chain information on the prediction remains to be explored. We evaluated the performance of multiple embedding models to predict BCR sequence properties and receptor specificity. Despite the differences in model architectures, most embeddings effectively capture BCR sequence properties and specificity. BCR-specific embeddings slightly outperform general protein language models in predicting specificity. In addition, incorporating full-length heavy chains and paired light chain sequences improve the prediction performance of all embeddings. This study provides insights into the properties of BCR embeddings to improve downstream prediction applications for antibody analysis and discovery.

bioinformatics↗

bioRxiv2023

The memory of pathogenic IgE is contained within CD23+IgG1+ memory B cells poised to switch to IgE in food allergy

Food allergy is caused by allergen-specific IgE antibodies but little is known about the B cell memory of persistent IgE responses. Here we describe in human pediatric peanut allergy CD23+IgG1+ memory B cells arising in type 2 responses that contain peanut specific clones and generate IgE cells on activation. These type2-marked IgG1+ memory B cells differentially express IL-4/IL-13 regulated genes FCER2/CD23, IL4R, and germline IGHE and carry highly mutated B cell receptors (BCRs). Further, high affinity memory B cells specific for the main peanut allergen Ara h 2 mapped to the population of type2-marked IgG1+ memory B cells and included convergent BCRs across different individuals. Our findings indicate that CD23+IgG1+ memory B cells transcribing germline IGHE are a unique memory population containing precursors of pathogenic IgE. One-Sentence SummaryWe describe a unique population of IgG+ memory B cells poised to switch to IgE that contains high affinity allergen-specific clones in peanut allergy.

immunology↗

bioRxiv2022

Transcriptional atlas of the human immune response to 13 vaccines reveals a common predictor of vaccine-induced antibody responses

Systems biology approaches have been used to define molecular signatures and mechanisms of immunity to vaccination. However, most such studies have been done with single vaccines, and comparative analysis of the response to different vaccines is lacking. We integrated temporal transcriptional data of over 3,000 samples, obtained from 820 healthy adults across 28 studies of 13 different vaccines and analyzed vaccination-induced signatures associated with the antibody response. Most vaccines induced similar kinetics of shared transcriptional signatures, including signatures of innate immunity occurring 1-3 days post-vaccination, as well as the canonical plasmablast and cell cycle signatures appearing 7 days post-vaccination. However, the yellow fever vaccine YF-17D uniquely induced an early transient signature of T and B cell activation at Day 1, followed by delayed antiviral/interferon and plasmablast signatures that peaked at Days 7 and 14-21, respectively. Thus, despite the shared transcriptional response to most vaccines, at any given time point there was no evidence for a "universal signature" that could be used to predict the antibody response to all vaccines. However, accounting for the asynchronous nature of responses led to the identification of a time-adjusted signature that improved prediction antibody of responses across vaccines. These results provide a transcriptional atlas of the human immune response to vaccination and define a common, time-adjusted signature of antibody responses to vaccination.

immunology↗

bioRxiv2021

The Immune Signatures Data Resource: A compendium of systems vaccinology datasets

Vaccines are among the most cost-effective public health interventions for preventing infection-induced morbidity and mortality, yet much remains to be learned regarding the mechanisms by which vaccines protect. Systems immunology combines traditional immunology with modern omic profiling techniques and computational modeling to promote rapid and transformative advances in vaccinology and vaccine discovery. The NIH/NIAID Human Immunology Project Consortium (HIPC) has leveraged systems immunology approaches to identify molecular signatures associated with the immunogenicity of many vaccines, including those targeting seasonal influenza, yellow fever, and hepatitis B. These data are made available to the broader scientific community through the ImmuneSpace data portal and analysis engine leveraging the NIH/NIAID ImmPort repository1,2. However, a barrier to progress in this area is that comparative analyses have been limited by the distributed nature of some data, potential batch effects across studies, and the absence of multiple relevant studies from non-HIPC groups in ImmPort. To support comparative analyses across different vaccines, we have created the Immune Signatures Data Resource, a compendium of standardized systems vaccinology datasets. This data resource is available through ImmuneSpace, along with code to reproduce the processing and batch normalization starting from the underlying study data in ImmPort and the Gene Expression Omnibus (GEO). The current release comprises 1405 participants from 53 cohorts profiling the response to 24 different vaccines and includes transcriptional profiles and antibody response measurements. This novel systems vaccinology data release represents a valuable resource for comparative and meta-analyses that will accelerate our understanding of mechanisms underlying vaccine responses. [Table 1]

systems biology↗

bioRxiv2020

Proteomics on longitudinal serum samples differentiates localized and disseminated Lyme disease

Lyme disease (LD) is the most prevalent vector-borne disease in North America, with ~300,000 new cases annually in the USA alone. LD is most often recognized by the appearance of the skin lesion erythema migrans (EM) at the tick bite site but also can present with signs of disseminated infection manifesting as multiple EM lesions and/or involvement of the heart, nervous system or joints. In this study, we examined the serum proteome of study participants presenting with either a single EM (localized LD) or early disseminated infection (disseminated LD). Samples collected at the time of diagnosis and at convalescent time points were assessed using our in-house developed MStern blotting-based serum proteomics platform. After technical validation of our platform, the temporal analysis from diagnosis to clinical resolution of the infection demonstrates LD stage-associated pathways activation such as a temporary upregulation of acute phase response specific to the participants with disseminated LD. In addition, we identified the members of the serum amyloid A protein family as potentially promising candidate biomarkers to identify those with disseminated LD. The results of this pilot study demonstrate the feasibility of using our time- and cost-effective sample sparing MStern blotting-based serum proteomics platform to efficiently interrogate proteome changes over time in those suffering infections such as LD. These observations establish a new approach to human serum proteomics, provide fresh insight into the host immune responses associated with disease severity (localized versus disseminated infection) and suggest novel biomarker candidate panels for LD stages. ImportanceWe investigated the proteome changes of Borrelia burgdorferi-infected participants with either a single erythema migrans or early disseminated Lyme disease infection. Using our in-house time-and cost-effective proteomics platform, the temporal analysis from diagnosis to clinical resolution of the infection shows a temporary upregulation of the acute phase response specific to the participants with disseminated infection. Finally, specific protein panels were identified as possible biomarker candidates to categorize those having an initial diagnosis of disseminated manifestation using a reference cohort of acute localized infection and a clinically resolved convalescent phase disease samples from the same Lyme disease participants.

immunology↗

bioRxiv2020

Immunologically distinct responses occur in the CNS of COVID-19 patients

One third of COVID-19 patients develop significant neurological symptoms, yet SARS-CoV-2 is rarely detected in central nervous system (CNS) tissue, suggesting a potential role for parainfectious processes, including neuroimmune responses. We therefore examined immune parameters in cerebrospinal fluid (CSF) and blood samples from a cohort of patients with COVID-19 and significant neurological complications. We found divergent immunological responses in the CNS compartment, including increased levels of IL-12 and IL-12-associated innate and adaptive immune cell activation. Moreover, we found increased proportions of B cells in the CSF relative to the periphery and evidence of clonal expansion of CSF B cells, suggesting a divergent intrathecal humoral response to SARS-CoV-2. Indeed, all COVID-19 cases examined had anti-SARS-CoV-2 IgG antibodies in the CSF whose target epitopes diverged from serum antibodies. We directly examined whether CSF resident antibodies target self-antigens and found a significant burden of CNS autoimmunity, with the CSF from most patients recognizing neural self-antigens. Finally, we produced a panel of monoclonal antibodies from patients CSF and show that these target both anti-viral and anti-neural antigens--including one mAb specific for the spike protein that also recognizes neural tissue. This exploratory immune survey reveals evidence of a compartmentalized and self-reactive immune response in the CNS meriting a more systematic evaluation of neurologically impaired COVID-19 patients. One Sentence SummaryA subset of COVID-19 patients with neurologic impairment show cerebrospinal fluid-specific immune alterations that point to both neuroinvasion and anti-neural autoimmunity as potential causes of impairment.

immunology↗

bioRxiv2020

Alignment free identification of clones in B cell receptor repertoires

Following pathogenic challenge, activated B cells rapidly expand and undergo somatic hypermutation, yielding groups of clonally related B-cells with diversified immunoglobulin receptors. Inference of clonal relationships based on the receptor sequence is an essential step in many adaptive immune receptor repertoire sequencing studies. These relationships are typically identified by a multi-step process that involves: (1) grouping sequences based on shared V and J gene assignments, and junction lengths, and (2) clustering these sequences using a junction-based distance. However, this approach is sensitive to the initial V(D)J gene assignments, which are error-prone, and fails to identify clonal relatives whose junction length has changed through accumulation of indels. Through defining a translation-invariant feature space in which we cluster the sequences, we develop an alignment-free clonal identification method that does not require gene assignments and is not restricted to a fixed junction length. This alignment-free approach has higher sensitivity compared to a typical junction-based distance method without loss of specificity and PPV. While the alignment-free procedure identifies clones that are broadly consistent with the junction-based distance method, it also identifies clones with characteristics (multiple V or J gene assignments or junction lengths) that are not detectable with the junction based distance method.

immunology↗

bioRxiv2019

Single-cell immune repertoire tracing identifies rituximab refractory B cells that emerge during relapse

Rituximab, a B cell-depleting therapy, is indicated for treating a growing number of autoantibody-mediated autoimmune disorders. However, relapses can occur after treatment and autoantibody-producing B cell subsets may be found during relapses. It is not understood if these autoantibody-producing B cell subsets emerge from the failed depletion of pre-existing B cells or are re-generated de novo. To further define the mechanisms that cause post-rituximab relapse, we studied patients with autoantibody-mediated muscle-specific kinase (MuSK) myasthenia gravis (MG) who relapsed after treatment. We carried out single-cell transcriptional and B cell receptor (BCR) profiling on longitudinal B cell samples. We identified clones present prior to therapy that continued to persist during relapse. Persistent B cell clones included both antibody-secreting cells and memory B cells characterized by gene expression signatures associated with B cell survival. A subset of persistent antibody-secreting cells and memory B cells were specific for the MuSK autoantigen. These results demonstrate that rituximab is not fully effective at eliminating autoantibody-producing B cells and provide a mechanistic understanding of post-rituximab relapse in MuSK MG.

immunology↗

bioRxiv2019

Somatic hypermutation analysis for improved identification of B cell clonal families from next-generation sequencing data

MotivationAdaptive immune receptor repertoire sequencing (AIRR-Seq) offers the possibility of identifying and tracking B cell clonal expansions during adaptive immune responses. Members of a B cell clone are descended from a common ancestor and share the same initial V(D)J rearrangement, but their B cell receptore (BCR) sequence may differ due to the accumulation of somatic hypermutations (SHMs). Clonal relationships are learned from AIRR-seq data by analyzing the BCR sequence, with the most common methods focused on the highly diverse junction region. However, clonally related cells often share SHMs which have been accumulated during affinity maturation. Here, we investigate whether shared SHMs in the V and J segments of the BCR can be leveraged along with the junction sequence to improve the ability to identify clonally related sequences. We develop independent distance functions that capture junction similarity and shared mutations, and combine these in a spectral clustering framework to infer the BCR clonal relationships. Using both simulated and experimental data, we show that this model improves both the sensitivity and specificity for identifying B cell clones. AvailabilitySource code for this method is freely available in the SCOPer (Spectral Clustering for clOne Partitioning) R package (version 0.2 or later) in the Immcantation framework: www.immcantation.org under the CC BY-SA 4.0 license. Contactsteven.kleinstein@yale.edu

immunology↗

bioRxiv2019

Seasonal Variability and Shared Molecular Signatures of Inactivated Influenza Vaccination in Young and Older Adults

The seasonal influenza vaccine is an important public health tool but is only effective in a subset of individuals. The identification of molecular signatures provides a mechanism to understand the drivers of vaccine-induced immunity. Most previously reported molecular signatures of influenza vaccination were derived from a single age group or season, ignoring the effects of immunosenescence or vaccine composition. Thus, it remains unclear how immune signatures of vaccine response change with age across multiple seasons. Here we profile the transcriptional landscape of young and older adults over five consecutive vaccination seasons to identify shared signatures of vaccine response as well as marked seasonal differences. Along with substantial variability in vaccine-induced signatures across seasons, we uncovered a common transcriptional signature 28 days post-vaccination in both young and older adults. However, gene expression patterns associated with vaccine-induced antibody responses were distinct in young and older adults; for example, increased expression of Killer Cell Lectin Like Receptor B1 (KLRB1; CD161) 28 days post-vaccination positively and negatively predicted vaccine-induced antibody responses in young and older adults, respectively. These findings contribute new insights for developing more effective influenza vaccines, particularly in older adults.

immunology↗

bioRxiv2019

Repertoire-wide phylogenetic models of B cell molecular evolution reveal evolutionary signatures of aging and vaccination

In order to produce effective antibodies, B cells undergo rapid somatic hypermutation (SHM) and selection for binding affinity to antigen via a process called affinity maturation. The similarities between this process and evolution by natural selection have led many groups to use evolutionary and phylogenetic methods to characterize the development of immunological memory, vaccination, and other processes that depend on affinity maturation. However, these applications are limited by several features of affinity maturation that violate assumptions in standard phylogenetic models. Further, most phylogenetic models are designed to be applied to individual lineages comprising genetically diverse sequences, while B cell repertoires often consist of hundreds to thousands of separate low-diversity lineages. Here, we introduce a hierarchical phylogenetic framework that incorporates the unique features of SHM, and integrates information from all lineages in a repertoire to more precisely estimate model parameters. We demonstrate the power of this approach by characterizing previously un-described phenomena in affinity maturation. First, we find evidence consistent with age related changes in SHM hot-and cold-spot motifs. Second, we identify a consistent relationship between increased tree length and signs of increased negative selection, apparent in the repertoires of both healthy subjects and those undergoing active immune responses. This suggests that B cell lineages shift towards negative selection over time as a general feature of affinity maturation. Our study provides a framework for undertaking repertoire-wide phylogenetic testing of SHM hypotheses, and provides a new means of charactering the process of mutation and selection during affinity maturation.

immunology↗