Here, we present Lancet2, an open-source somatic variant caller designed to improve detection of small variants in short-read sequencing data. Lancet2 introduces significant enhancements, including: (i) Improved variant discovery and genotyping through partial order multiple sequence alignment of assembled haplotype contigs, and re-alignment of sample reads to the best supporting allele. (ii) Optimized somatic variant scoring with explainable machine learning models, leading to better somatic filtering throughout the sensitivity scale. (iii) Integration with Sequence Tube Map for enhanced visualization of variants with aligned sample reads in graph space. When benchmarked against enhanced two-tech truth sets generated using high-coverage short-read (Illumina) and long-read (Oxford Nanopore) data from four well characterized matched tumor/normal cell lines, Lancet2 outperformed other industry-leading tools in variant calling performance, especially for InDels. In addition, significant runtime performance improvements were observed compared to Lancet1 (∼10× speedup and 50% less peak memory usage), and most other state-of-the-art somatic variant callers (at least 2× speedup with eight cores or more), making Lancet2 an ideal tool for accurate and efficient somatic variant calling.

Gene expression quantification through genomics methods is crucial for understanding diverse biological contexts. Among these methods, ribosome profiling (Ribo-seq) stands out as a valuable tool for uncovering post-transcriptional gene expression regulation by providing a comprehensive view of the translatome. While current protocols are time-intensive with limited variations, we introduced the use of the Benzonase enzyme to generate ribosome footprints from a polysome-enriched fraction, streamlining the workflow and reducing time, cost and potential sources of bias. Comparing translatome from Benzonase- and RNAse I-derived footprints reveals minimal differences, with both separating from RNA-seq-based transcriptome quantification. Benzonase acts more gently on ribosomes, yielding larger footprints and a weaker triplet periodicity compared to RNAse I, while enabling its application to low-input samples. We further demonstrate Ribo-seq application in primary neuronal cultures, using Benzonase to digest the post-mitochondrial supernatant, thereby bypassing the labor-intensive ribosome/polysome purification step. The introduction of such protocol variations for Ribo-seq, especially for challenging low-input samples, offers a significant advancement of this resource for the community.

Although an increasing number of long-read genome assemblies have been created from a diverse collection of dogs and wolves, most published assemblies represent the diploid genome as a single primary sequence. Here, we generate and analyze phase-resolved diploid dual assemblies from five canines. The most contiguous assemblies represent over half of the canine chromosomes as single contigs, permitting an assessment of the sequence and structure of canine chromosomes. Consistent with a telocentric classification, we find that the centromeres of canine autosomes begin an average of 59 kb from the start of the chromosome and are flanked by a 35 kb subtelomeric segment that is repeat-rich and shared across autosomes. Analysis of a pangenome graph constructed from the 10 haplotype-resolved assemblies shows that short tandem repeat loci are three times more common than variable number tandem repeat loci and that the landscape of canine structural variation features extensive allelic heterogeneity. The pangenome graph includes examples of complex, nested allelic variation involving SINEC (a carnivore-specific SINE) and LINE-1 mobile elements. Analysis of 3′ transductions implicate an uncharacterized source element with high activity and demonstrates the presence of full-length LINE-1s capable of retrotransposition that are segregating among canines.

Large Language Models (LLMs) are increasingly applied to genomic tasks, yet core challenges remain concerning tokenization, evaluation, and data scarcity. This study focuses on promoter classification and systematically evaluates four tokenization methods: non-overlapping 6-mer, overlapping 6-mer, Byte Pair Encoding (BPE), and WordPiece (WPC). We show that the commonly used k-mer approach, specifically the non-overlapping variant, outperforms BPE and WPC across eight organisms, challenging assumptions derived from natural language processing. To ensure robustness, we evaluated performance under two distinct negative data strategies: positive-promoter-shuffled and random-non-promoter-fragments. Using a positional SHAP framework, we demonstrate that the model learns biologically plausible positional patterns rather than exploiting artifacts from these negative data generation processes. Furthermore, evolutionary-informed transfer learning experiments and external validation on an unseen organism reveal that training on phylogenetically related species significantly improves performance, particularly in low-data regimes. These findings underscore the significant impact of tokenization and negative data design, providing practical guidance for refining genomic classifiers.

Chromatin conformation capture methods such as Hi-C have improved understanding of nuclear architecture. However, reconstruction from single-cell Hi-C (scHi-C) data is challenging due to limited DNA contacts per cell. We have previously developed the recurrence plot-based reconstruction (RPR) method for reconstructing three-dimensional (3D) genomic structure from Hi-C data even from low-coverage DNA contact information. Here we used the RPR method to analyze scHi-C data derived from early-stage F1 hybrid embryos as a proof-of-concept for understanding of global chromosomal architecture. We found that paternal and maternal chromosomes become gradually intermingled from the 1-cell to the 64-cell stage, and that discrete chromosome territories are largely established between 8-cell and 64-cell stages. We also observed Rabl-like polarization of chromosomes from the 2- to 8-cell stage, which was mostly dissolved by the 64-cell stage. We also noted transient rod-like extension and parallel chromosome alignment at the 4-cell stage. These findings indicate dynamic chromosomal reorganization before territory formation. RPR and scHi-C together capture 3D chromosomal architecture of individual cells during early embryogenesis.

In this study, we developed an integrated single-cell transcriptomic (scRNAseq) atlas of human breast cancer (BC), the largest resource of its kind, totaling >600 000 cells across 138 patients. Rigorous integration and annotation of publicly available scRNAseq data enabled a highly resolved characterization of epithelial, immune, and stromal heterogeneity within the tumor microenvironment (TME). Within the immune compartment, we were able to characterize heterogeneity of CD4, CD8 T cells, and macrophage subpopulations. Within the stromal compartment, subpopulations of endothelial cells (ECs) and cancer-associated fibroblasts (CAFs) were resolved. Within the cancer epithelial compartment, we characterized the functional heterogeneity of cells across the axes of stemness, epithelial-mesenchymal plasticity, and canonical cancer pathways. Across all subpopulations observed in the TME, we performed a multi-resolution survival analysis to identify epithelial cell states and immune and stromal cell types, which conferred a survival advantage in both The Cancer Genome Atlas (TCGA), METABRIC, and SCANB. We also identified robust associations between TME composition and clinical phenotypes such as tumor subtype and grade that were not discernible when the analysis was limited to individual datasets, highlighting the need for atlas-based analyses. This atlas represents a valuable resource for further high-resolution analyses of TME heterogeneity within BC.

Here, we introduce a new annotation pipeline, called Eukan, designed to deliver reliably high-quality results across a broad range of eukaryotes. First, experimental evidence is automatically leveraged to refine predictions, specifically, RNA-Seq coverage to inform generalized Hidden Markov Model gene prediction and intron lengths to inform protein sequence alignments. Second, a consensus is created from an empirically optimized weighting of gene predictions from multiple sources. Third, Eukan runs a post-annotation routine to recover gene predictions missing from the consensus that otherwise have strong transcript support and appear to be protein-coding. We compare the results of Eukan with those of three popular freely available pipelines (Maker, Braker, and Gemoma) on 17 phylogenetically diverse haploid and diploid nuclear genomes. In addition to the commonly reported annotation accuracy statistics, we define a novel classification system of critical defects commonly observed in automated annotations. Furthermore, we demonstrate that each of the tested pipelines correctly identified the majority of the validated “gold standard” genes across the test set, but each pipeline uniquely generates a non-negligible portion of either fragmented, artificially fused, or missing genes. Despite that, Eukan performs consistently well where other pipelines encounter challenges, such as for compact protist genomes.

Pneumocystis jirovecii poses a significant threat to immunocompromised individuals, necessitating the development of an effective vaccine. This study employs an immunoinformatics approach to design a promising vaccine candidate against P. jirovecii. Utilizing various computational tools, the study identified potential antigenic epitopes capable of eliciting immune responses within the P. jirovecii major surface glycoprotein C. The chosen epitopes were evaluated using computational tools for their allergenicity, interferon-γ and interleukin activation ability, and toxicity, ensuring the selection of immunogenic and safe candidates. These analyses led to the selection of 10 epitopes, which were then linked with adjuvants to model a potential vaccine candidate. Molecular docking and molecular dynamics simulations were performed in a solvent environment to investigate the binding interactions between the vaccine candidate and toll-like receptors, along with calculations of thermodynamic properties. Finally, in silico immune simulations were performed to analyze the immunogenic potential of the vaccine candidate. Future prospects include in vitro and in vivo validation of the vaccine candidate and the exploration of novel adjuvants to enhance its immunogenicity. This study contributes to the ongoing efforts to develop a preventive solution against P. jirovecii infections, addressing a critical gap in the protection of immunocompromised individuals.

Genomics research increasingly relies on large population biobanks that include many thousands of participants. However, current genetic ancestry inference methods are computationally inefficient and prohibitively slow when applied to such large cohorts. The aim of this work was to develop a fast and efficient algorithm for fine-scale genetic ancestry inference on biobank-size cohorts. The Pastrami algorithm that we developed performs supervised genetic ancestry inference by comparing haplotypes between query and global reference samples, creating query and reference haplotype copying vectors, and relating them via non-negative least squares regression to estimate ancestry fractions. We used Pastrami for ancestry inference on genomic data sets from Africa, the Americas, and the United Kingdom, comparing its accuracy and runtime performance to the most widely used haplotype-based ancestry inference methods. Pastrami ancestry estimates are highly similar to estimates from the ChromoPainter and RFMix programs. The total CPU time required by Pastrami increases linearly with the number of samples, and it achieves ∼45× faster runtime than ChromoPainter. When run on 488 377 UK Biobank and 3433 reference samples, Pastrami used 2340 CPU hours compared to ∼105 000 CPU hours for ChromoPainter. The Pastrami program and documentation are made freely available on GitHub: https://github.com/healthdisparities/pastrami.