Precision Neoantigen Discovery Using Large-Scale Immunopeptidomes and Composite Modeling of MHC Peptide Presentation.

Major histocompatibility complex (MHC)-bound peptides that originate from tumor-specific genetic alterations, known as neoantigens, are an important class of anticancer therapeutic targets. Accurately predicting peptide presentation by MHC complexes is a key aspect of discovering therapeutically relevant neoantigens. Technological improvements in mass spectrometry-based immunopeptidomics and advanced modeling techniques have vastly improved MHC presentation prediction over the past 2 decades. However, improvement in the accuracy of prediction algorithms is needed for clinical applications like the development of personalized cancer vaccines, the discovery of biomarkers for response to immunotherapies, and the quantification of autoimmune risk in gene therapies. Toward this end, we generated allele-specific immunopeptidomics data using 25 monoallelic cell lines and created Systematic Human Leukocyte Antigen (HLA) Epitope Ranking Pan Algorithm (SHERPA), a pan-allelic MHC-peptide algorithm for predicting MHC-peptide binding and presentation. In contrast to previously published large-scale monoallelic data, we used an HLA-null K562 parental cell line and a stable transfection of HLA allele to better emulate native presentation. Our dataset includes five previously unprofiled alleles that expand MHC diversity in the training data and extend allelic coverage in underprofiled populations. To improve generalizability, SHERPA systematically integrates 128 monoallelic and 384 multiallelic samples with publicly available immunoproteomics data and binding assay data. Using this dataset, we developed two features that empirically estimate the propensities of genes and specific regions within gene bodies to engender immunopeptides to represent antigen processing. Using a composite model constructed with gradient boosting decision trees, multiallelic deconvolution, and 2.15 million peptides encompassing 167 alleles, we achieved a 1.44-fold improvement of positive predictive value compared with existing tools when evaluated on independent monoallelic datasets and a 1.17-fold improvement when evaluating on tumor samples. With a high degree of accuracy, SHERPA has the potential to enable precision neoantigen discovery for future clinical applications.

A machine learning algorithm with subclonal sensitivity reveals widespread pan-cancer human leukocyte antigen loss of heterozygosity.

Human leukocyte antigen loss of heterozygosity (HLA LOH) allows cancer cells to escape immune recognition by deleting HLA alleles, causing the suppressed presentation of tumor neoantigens. Despite its importance in immunotherapy response, few methods exist to detect HLA LOH, and their accuracy is not well understood. Here, we develop DASH (Deletion of Allele-Specific HLAs), a machine learning-based algorithm to detect HLA LOH from paired tumor-normal sequencing data. With cell line mixtures, we demonstrate increased sensitivity compared to previously published tools. Moreover, our patient-specific digital PCR validation approach provides a sensitive, robust orthogonal approach that could be used for clinical validation. Using DASH on 610 patients across 15 tumor types, we find that 18% of patients have HLA LOH. Moreover, we show inflated HLA LOH rates compared to genome-wide LOH and correlations between CD274 (encodes PD-L1) expression and microsatellite instability status, suggesting the HLA LOH is a key immune resistance strategy.

Prediction of Immunotherapy Response in Melanoma through Combined Modeling of Neoantigen Burden and Immune-Related Resistance Mechanisms.

PURPOSE: While immune checkpoint blockade (ICB) has become a pillar of cancer treatment, biomarkers that consistently predict patient response remain elusive due to the complex mechanisms driving immune response to tumors. We hypothesized that a multi-dimensional approach modeling both tumor and immune-related molecular mechanisms would better predict ICB response than simpler mutation-focused biomarkers, such as tumor mutational burden (TMB). EXPERIMENTAL DESIGN: Tumors from a cohort of patients with late-stage melanoma (n = 51) were profiled using an immune-enhanced exome and transcriptome platform. We demonstrate increasing predictive power with deeper modeling of neoantigens and immune-related resistance mechanisms to ICB. RESULTS: Our neoantigen burden score, which integrates both exome and transcriptome features, more significantly stratified responders and nonresponders (P = 0.016) than TMB alone (P = 0.049). Extension of this model to include immune-related resistance mechanisms affecting the antigen presentation machinery, such as HLA allele-specific LOH, resulted in a composite neoantigen presentation score (NEOPS) that demonstrated further increased association with therapy response (P = 0.002). CONCLUSIONS: NEOPS proved the statistically strongest biomarker compared with all single-gene biomarkers, expression signatures, and TMB biomarkers evaluated in this cohort. Subsequent confirmation of these findings in an independent cohort of patients (n = 110) suggests that NEOPS is a robust, novel biomarker of ICB response in melanoma.

Precision Neoantigen Discovery Using Large-scale Immunopeptidomes and Composite Modeling of MHC Peptide Presentation.

Major histocompatibility complex (MHC)-bound peptides that originate from tumor-specific genetic alterations, known as neoantigens, are an important class of anticancer therapeutic targets. Accurately predicting peptide presentation by MHC complexes is a key aspect of discovering therapeutically relevant neoantigens. Technological improvements in mass-spectrometry-based immunopeptidomics and advanced modeling techniques have vastly improved MHC presentation prediction over the past two decades. However, improvement in the sensitivity and specificity of prediction algorithms is needed for clinical applications such as the development of personalized cancer vaccines, the discovery of biomarkers for response to checkpoint blockade, and the quantification of autoimmune risk in gene therapies. Toward this end, we generated allele-specific immunopeptidomics data using 25 monoallelic cell lines and created Systematic HLA Epitope Ranking Pan Algorithm (SHERPA), a pan-allelic MHC-peptide algorithm for predicting MHC-peptide binding and presentation. In contrast to previously published large-scale monoallelic data, we used an HLA-null K562 parental cell line and a stable transfection of HLA alleles to better emulate native presentation. Our dataset includes five previously unprofiled alleles that expand MHC-binding pocket diversity in the training data and extend allelic coverage in under profiled populations. To improve generalizability, SHERPA systematically integrates 128 monoallelic and 384 multiallelic samples with publicly available immunoproteomics data and binding assay data. Using this dataset, we developed two features that empirically estimate the propensities of genes and specific regions within gene bodies to engender immunopeptides to represent antigen processing. Using a composite model constructed with gradient boosting decision trees, multiallelic deconvolution, and 2.15 million peptides encompassing 167 alleles, we achieved a 1.44-fold improvement of positive predictive value compared with existing tools when evaluated on independent monoallelic datasets and a 1.15-fold improvement when evaluating on tumor samples. With a high degree of accuracy, SHERPA has the potential to enable precision neoantigen discovery for future clinical applications.

Intraspecific Genetic Variation Underlying Postmating Reproductive Barriers between Species in the Wild Tomato Clade (Solanum sect. Lycopersicon).

A goal of speciation genetics is to understand how the genetic components underlying interspecific reproductive barriers originate within species. Unilateral incompatibility (UI) is a postmating prezygotic barrier in which pollen rejection in the female reproductive tract (style) occurs in only one direction of an interspecific cross. Natural variation in the strength of UI has been observed among populations within species in the wild tomato clade. In some cases, molecular loci underlying self-incompatibility (SI) are associated with this variation in UI, but the mechanistic connection between these intra- and inter-specific pollen rejection behaviors is poorly understood in most instances. We generated an F2 population between SI and SC genotypes of a single species, Solanum pennellii, to examine the genetic basis of intraspecific variation in UI against other species, and to determine whether loci underlying SI are genetically associated with this variation. We found that F2 individuals vary in the rate at which UI rejection occurs. One large effect QTL detected for this trait co-localized with the SI-determining S-locus. Moreover, individuals that expressed S-RNase-the S-locus protein involved in SI pollen rejection-in their styles had much more rapid UI responses compared with those without S-RNase protein. Our analysis shows that intraspecific variation at mate choice loci-in this case at loci that prevent self-fertilization-can contribute to variation in the expression of interspecific isolation, including postmating prezygotic barriers. Understanding the nature of such intraspecific variation can provide insight into the accumulation of these barriers between diverging lineages.

Evolutionary superscaffolding and chromosome anchoring to improve Anopheles genome assemblies.

BACKGROUND: New sequencing technologies have lowered financial barriers to whole genome sequencing, but resulting assemblies are often fragmented and far from 'finished'. Updating multi-scaffold drafts to chromosome-level status can be achieved through experimental mapping or re-sequencing efforts. Avoiding the costs associated with such approaches, comparative genomic analysis of gene order conservation (synteny) to predict scaffold neighbours (adjacencies) offers a potentially useful complementary method for improving draft assemblies. RESULTS: We evaluated and employed 3 gene synteny-based methods applied to 21 Anopheles mosquito assemblies to produce consensus sets of scaffold adjacencies. For subsets of the assemblies, we integrated these with additional supporting data to confirm and complement the synteny-based adjacencies: 6 with physical mapping data that anchor scaffolds to chromosome locations, 13 with paired-end RNA sequencing (RNAseq) data, and 3 with new assemblies based on re-scaffolding or long-read data. Our combined analyses produced 20 new superscaffolded assemblies with improved contiguities: 7 for which assignments of non-anchored scaffolds to chromosome arms span more than 75% of the assemblies, and a further 7 with chromosome anchoring including an 88% anchored Anopheles arabiensis assembly and, respectively, 73% and 84% anchored assemblies with comprehensively updated cytogenetic photomaps for Anopheles funestus and Anopheles stephensi. CONCLUSIONS: Experimental data from probe mapping, RNAseq, or long-read technologies, where available, all contribute to successful upgrading of draft assemblies. Our evaluations show that gene synteny-based computational methods represent a valuable alternative or complementary approach. Our improved Anopheles reference assemblies highlight the utility of applying comparative genomics approaches to improve community genomic resources.

The sequencing and interpretation of the genome obtained from a Serbian individual.

Recent genetic studies and whole-genome sequencing projects have greatly improved our understanding of human variation and clinically actionable genetic information. Smaller ethnic populations, however, remain underrepresented in both individual and large-scale sequencing efforts and hence present an opportunity to discover new variants of biomedical and demographic significance. This report describes the sequencing and analysis of a genome obtained from an individual of Serbian origin, introducing tens of thousands of previously unknown variants to the currently available pool. Ancestry analysis places this individual in close proximity to Central and Eastern European populations; i.e., closest to Croatian, Bulgarian and Hungarian individuals and, in terms of other Europeans, furthest from Ashkenazi Jewish, Spanish, Sicilian and Baltic individuals. Our analysis confirmed gene flow between Neanderthal and ancestral pan-European populations, with similar contributions to the Serbian genome as those observed in other European groups. Finally, to assess the burden of potentially disease-causing/clinically relevant variation in the sequenced genome, we utilized manually curated genotype-phenotype association databases and variant-effect predictors. We identified several variants that have previously been associated with severe early-onset disease that is not evident in the proband, as well as putatively impactful variants that could yet prove to be clinically relevant to the proband over the next decades. The presence of numerous private and low-frequency variants, along with the observed and predicted disease-causing mutations in this genome, exemplify some of the global challenges of genome interpretation, especially in the context of under-studied ethnic groups.

AGOUTI: improving genome assembly and annotation using transcriptome data.

BACKGROUND: Genomes sequenced using short-read, next-generation sequencing technologies can have many errors and may be fragmented into thousands of small contigs. These incomplete and fragmented assemblies lead to errors in gene identification, such that single genes spread across multiple contigs are annotated as separate gene models. Such biases can confound inferences about the number and identity of genes within species, as well as gene gain and loss between species. RESULTS: We present AGOUTI (Annotated Genome Optimization Using Transcriptome Information), a tool that uses RNA sequencing data to simultaneously combine contigs into scaffolds and fragmented gene models into single models. We show that AGOUTI improves both the contiguity of genome assemblies and the accuracy of gene annotation, providing updated versions of each as output. Running AGOUTI on both simulated and real datasets, we show that it is highly accurate and that it achieves greater accuracy and contiguity when compared with other existing methods. CONCLUSION: AGOUTI is a powerful and effective scaffolder and, unlike most scaffolders, is expected to be more effective in larger genomes because of the commensurate increase in intron length. AGOUTI is able to scaffold thousands of contigs while simultaneously reducing the number of gene models by hundreds or thousands. The software is available free of charge under the MIT license.

Sequencing, assembling, and correcting draft genomes using recombinant populations.

Current de novo whole-genome sequencing approaches often are inadequate for organisms lacking substantial preexisting genetic data. Problems with these methods are manifest as: large numbers of scaffolds that are not ordered within chromosomes or assigned to individual chromosomes, misassembly of allelic sequences as separate loci when the individual(s) being sequenced are heterozygous, and the collapse of recently duplicated sequences into a single locus, regardless of levels of heterozygosity. Here we propose a new approach for producing de novo whole-genome sequences-which we call recombinant population genome construction-that solves many of the problems encountered in standard genome assembly and that can be applied in model and nonmodel organisms. Our approach takes advantage of next-generation sequencing technologies to simultaneously barcode and sequence a large number of individuals from a recombinant population. The sequences of all recombinants can be combined to create an initial de novo assembly, followed by the use of individual recombinant genotypes to correct assembly splitting/collapsing and to order and orient scaffolds within linkage groups. Recombinant population genome construction can rapidly accelerate the transformation of nonmodel species into genome-enabled systems by simultaneously producing a high-quality genome assembly and providing genomic tools (e.g., high-confidence single-nucleotide polymorphisms) for immediate applications. In populations segregating for important functional traits, this approach also enables simultaneous mapping of quantitative trait loci. We demonstrate our method using simulated Illumina data from a recombinant population of Caenorhabditis elegans and show that the method can produce a high-fidelity, high-quality genome assembly for both parents of the cross.