Biology subjects

Genomics

Explore bioRxiv genomics preprints on sequencing, gene expression and genetic variation. Read source abstracts and verify methods in the manuscript; preprints are not peer reviewed.

Follow this topic with RSS

At least 361 records · Page 20Linked to original sources

bioRxiv2021

Capturing single-copy nuclear genes, organellar genomes, and nuclear ribosomal DNA from deep genome skimming data for plant phylogenetics: A case study in Vitaceae

With the decreasing cost and availability of many newly developed bioinformatics pipelines, next-generation sequencing (NGS) has revolutionized plant systematics in recent years. Genome skimming has been widely used to obtain high-copy fractions of the genomes, including plastomes, mitochondrial DNA (mtDNA), and nuclear ribosomal DNA (nrDNA). In this study, through simulations, we evaluated optimal (minimum) sequencing depth and performance for recovering single-copy nuclear genes (SCNs) from genome skimming data, by subsampling genome resequencing data and generating 10 datasets with different sequencing coverage in silico. We tested the performance of the four datasets (plastome, nrDNA, mtDNA, and SCNs) obtained from genome skimming based on phylogenetic analyses of the Vitis clade at the genus-level and Vitaceae at the family-level, respectively. Our results showed that optimal minimum sequencing depth for high-quality SCNs assembly via genome skimming was about 10x coverage. Without the steps of synthesizing baits and enrichment experiments, we showcase that deep genome skimming (DGS) is effective for capturing large datasets of SCNs, in addition to plastomes, mtDNA, and entire nrDNA repeats, and may serve as an economical alternative to the widely used target enrichment Hyb-Seq approach.

evolutionary biology↗

bioRxiv2021

Pleiotropic Variability Score: A Genome Interpretation Metric to Quantify Phenomic Associations of Genomic Variants

A more complete understanding of phenomic space is critical for elucidating genome-phenome relationships and for assessing disease risk from genome sequencing. To incorporate knowledge of how related a variants associations are, we developed a new genome interpretation metric called Pleiotropic Variability Score (PVS). PVS uses semantic reasoning to score the relatedness of a genetic variants associated phenotypes based on those phenotypes relationships in the human phenotype ontology (HPO) and disease ontology (DO). We tested 78 unique semantic similarity methods and integrated six robust metrics to define the pleiotropy score of SNPs. We computed PVS for 12,541 SNPs which were mapped to 382 HPO and 317 DO unique phenotype terms in a genotype-phenotype catalog (10,021 SNPs mapped to DO phenotypes and 8,569 SNPs mapped to HPO phenotypes). We validated the utility of PVS by computing pleiotropy using an electronic health record linked genomic database (BioME, n=11,210). Further we demonstrate the application of PVS in personalized medicine using "personalized pleiotropy score" reports for individuals with genomic data that could potentially aid in variant interpretation. We further developed a software framework to incorporate PVS into VCF files and to consolidate pleiotropy assessment as part of genome interpretation pipelines. As the genome-phenome catalogs are growing, PVS will be a useful metric to assess genetic variation to find SNPs with highly pleiotropic effects. Additionally, variants with varying degree of pleiotropy can be prioritized for explorative studies to understand specific roles of SNPs and pleiotropic hubs in mediating novel phenotypes and drug development.

bioinformatics↗

bioRxiv2021

Gigantic Genomes of Salamanders Indicate Body Temperature, not Genome Size, is the Driver of Global Methylation and 5-Methylcytosine Deamination in Vertebrates

Methylation of cytosines at CpG dinucleotide sites silences transposable elements (TEs), sequences that replicate and move throughout genomes. TE abundance drives differences in genome size, but TE silencing variation across genomes of different sizes remains largely unexplored. Salamanders include most of the largest C-values -- 9 to 120 Gb. We measured CpG methylation levels in salamanders with genomes ranging from 2N = ~58 Gb to 4N = ~116 Gb. We compared these levels to results from endo- and ectothermic vertebrates with more typical genomes. Salamander methylation levels are ~90%, higher than all endotherms. However, salamander methylation does not differ from the other ectotherms, despite a ~100-fold difference in nuclear DNA content. Because methylation affects the nucleotide compositional landscape through 5-methylcytosine deamination to thymine, we quantified salamander CpG dinucleotide levels and compared them to other vertebrates. Salamanders have comparable CpG levels to other ectotherms, and ectotherm levels are higher than endotherms. These data show no shift in global methylation at the base of salamanders, despite a dramatic increase in TE load and genome size. This result is reconcilable with previous studies by considering endothermy and ectothermy, which may be more important drivers of methylation in vertebrates than genome size.

evolutionary biology↗

bioRxiv2022

Exploring Genomic Conservation in Actinobacteriophages with Small Genomes

Actinobacteriophages of a wide range of genome sizes continue to be isolated and characterized, but only a handful of these have atypically small genomes, defined in this work as genome sizes under 20,000 bp. These "small phages" are relatively rare and have received minimal study thus far. Of the actinobacteriophages published in PhagesDB.org, small phages have been isolated on Arthrobacter, Gordonia, Rhodococcus, and Microbacterium hosts. A previous study by Pope et al. showed that Gordonia small phages have similar gene products and amino acid sequences. Here, we set out to further examine relationships between small Gordonia phages as well as small phages that infect other hosts. Of the 3222 sequenced phages listed on PhagesDB, we identified 109 distinctly small phages with genome sizes under 20,000 bp. The majority of the small phages were isolated on Arthrobacter or Microbacterium hosts. Using comparative genomics, we searched for patterns of similarity among 34 cluster-representative small phages. Dot plot comparisons showed that there was more amino acid conservation than nucleotide identity amongst small phages. Gene content similarity (GCS) analysis revealed that the temperate Gordonia phages in Cluster CW share significant GCS values (over 35%) with the lytic Arthrobacter phages in Cluster AN, suggesting that some small phages have a considerable degree of genomic similarity with each other. SplitsTree analyses of shared phams (genes with substantial amino acid identity) supported the complexity of clustering criteria in small phages, given shuffling of genes across phages of different clusters and close relationships despite varied cluster membership. We observed this continuum of phage diversity through Rhodococcus phage RRH1s closer similarity to phages in Gordonia subcluster CW1 than CW1 is to Gordonia subcluster CW3. Finally, we were able to confirm the presence of conserved phams across not only small Gordonia phages but also within small phages from different clusters and hosts. Studying these genomic trends hidden in small phages allows us to better understand and appreciate the overall diversity of phages.

microbiology↗

bioRxiv2022

Insights into the genomics of clownfish adaptive radiation: the genomic substrate of the diversification

Clownfishes are an iconic group of coral reef fishes that evolved a mutualistic interaction with sea anemones, which triggered the rapid diversification of the group. We investigated the genomic architecture underlying this process to determine the genomic characteristics associated with the adaptive radiation of the group and assess the mechanisms of parallel evolution in clownfishes. We took advantage of the available genomic data of five pairs of closely related but ecologically divergent clownfish species to perform comparative genomic analyses. We found that clownfish genomes show two bursts of transposable elements, overall accelerated coding evolution, and topology inconsistencies potentially resulting from hybridization events. These characteristics possibly facilitated the rapid diversification of the group. We also detected a signature of positive selection throughout the radiation in 5.4 % of the clownfish genes. Among them, five presented functions associated with social behavior and ecology. They could have potentially played a role in the evolution of size-based hierarchical social structure so particular to clownfishes. Finally, we found genes with patterns of either relaxation or intensification of purifying selection and signals of positive selection linked with clownfish ecological divergence, suggesting some level of parallel evolution during the diversification of the group. Altogether, these results provide the first insights into the genomic substrate of clownfish adaptive radiation. This work integrates the growing collection of studies investigating the genomic mechanisms governing species diversification, which brings us a step closer to understanding how biodiversity on Earth is created

evolutionary biology↗

bioRxiv2022

The genome of Gynandropsis gynandra provides insights into whole-genome duplications and the evolution of C4 photosynthesis in Cleomaceae

Gynandropsis gynandra (Cleomaceae) is a cosmopolitan leafy vegetable and medicinal plant, which has also been used as a model to study C4 photosynthesis due to its evolutionary proximity to Arabidopsis. Here, we present a high-quality genome sequence of G. gynandra, anchored onto 17 main super- scaffolds with a total length of 740 Mb, an N50 of 42 Mb and 30,933 well-supported gene models. The G. gynandra genome and previously released genomes of C3 relatives in the Cleomaceae and Brassicaceae make an excellent model for studying the role of genome evolution in the transition from C3 to C4 photosynthesis. We revealed that G. gynandra and its C3 relative Tarenaya hassleriana shared a whole-genome duplication event (Gg-), then an addition of a third genome (Th-, +1x) took place in T. hassleriana but not in G. gynandra. Analysis of syntenic copy number of C4 photosynthesis-related gene families indicates that G. gynandra generally retained more duplicated copies of these genes than C3 T. hassleriana, and also that the G. gynandra C4 genes might have been under positive selection pressure. Both whole-genome and single-gene duplication were found to contribute to the expansion of the aforementioned gene families in G. gynandra. Collectively, this study enhances our understanding of the impact of gene duplication and gene retention on the evolution of C4 photosynthesis in Cleomaceae.

plant biology↗

bioRxiv2022

Probabilistic 3D-modelling of genomes and genomic domains by integrating high-throughput imaging and Hi-C using machine learning

Among the existing techniques for interrogating the genome structure, Hi-C assays have become the most performed experiments and constitute the majority of the publicly available datasets. As a result, there is a continuous demand to create and improve algorithms and methods to assist the scientific community in the interpretation of Hi-C experimental data. Here we introduce probabilistic TADbit (pTADbit), a new approach that combines Deep Learning and restraint-based modelling to infer the three-dimensional (3D) structure of genome and genomic domains interrogated by Hi-C experiments. pTADbit uses thousands of microscopy-based distances between genomic loci to train a neural network model that aims at predicting the population distribution of the spatial distance between two genomic loci based solely on their Hi-C interaction frequency. pTADbit produces more accurate chromatin models compared to the original TADbit as well as other available 3D modeling methods, while drastically reducing the required computation time. The resulting ensemble of models not only agree consistently with independent measures obtained by imaging experiments but also better capture the heterogeneity of the cell population. The development of pTADbit lays the basis for the integration of data produced from high-throughput imaging assays into the 3D modelling genomes and genomic domains.

bioinformatics↗

bioRxiv2022

Taxonomic vs genomic fungi: contrasting evolutionary loss of protistan genomic heritage and emergence of fungal novelties

Fungi are among the most ecologically important heterotrophs that have radiated into most niches on Earth and fulfil key ecological services. However, despite intense interest in their origins, major genomic trends characterising the evolutionary route from a unicellular opisthokont ancestor to derived multicellular fungi remain poorly known. Here, we reconstructed gene family evolution across 123 genomes of fungi and relatives and show that a dominant trend in early fungal evolution has been the gradual shedding of protist genes and highly episodic innovation via gene duplication. We find that the gene content of early-diverging fungi is protist-like in many respects, owing to the conservation of protist genes in early fungi. While gene loss has been constant and gradual during early fungal evolution, our reconstructions show that gene innovation showed two peaks. Gene groups with the largest contribution to genomic change included extracellular proteins, transcription factors, as well as ones linked to the coordination of nutrient uptake with growth, highlighting the transition to a sessile osmotrophic feeding strategy and subsequent lifestyle evolution as important elements of early fungal evolution. Taken together, this work provided a highly resolved genome-wide catalogue of gene family changes across fungal evolution. This suggests that the genome of pre-fungal ancestors may have been transformed into the archetypal fungal genome by a combination of gradual gene loss, turnover and two large duplication events rather than by abrupt changes, and consequently, that the taxonomically defined fungal kingdom does not represent a genomically uniform assemblage of extant species characterized by diagnostic synapomorphies.

evolutionary biology↗

bioRxiv2023

GenBank genomics highlight the genomic features, genetic diversity and regulation of morphological, metabolic and disease-resistance traits in Nicotiana tabacum

Nicotiana tabacum is a model organism in plant molecular and pathogenic research and has significant potential in the production of biofuels and active pharmaceutical compounds in synthetic biology. Because of the large allotetraploid genome of tobacco, its genomic features, genetic diversity and genetic regulation of many complex traits remain unknown. In this study, we present a nearly complete chromosome-scale assembly of N. tabacum and provide evidence that homoeologous exchange between subgenomes and epigenetic remodelling are likely mechanisms of genome stabilization and subgenome coordination following polyploidization. By leveraging GenBank-scale sequencing and phenotyping data from 5196 lines, geography at the continent scale, rather than types assigned on the basis of curing crop practices, was found to be the most important correlate of genetic structure. Using 178 marker{square}trait associations detected in genome-wide association analysis, a reference genotype-to-phenotype map was built for 39 morphological, developmental, and disease-resistance traits. A novel gene, auxin response factor 9 (Arf9), associated with wider leaves after being knocked out, was fine-mapped to a single nucleotide polymorphism (SNP). This point mutation alters the translated amino acid from Ala203 to Pro203, likely preventing homodimer formation during DNA binding. Our analysis also revealed signatures of positive and polygenic selection for multiple traits during the process of selective breeding. Overall, this study demonstrated the power of leveraging GenBank genomics to gain insights into the genomic features, genetic diversity, and regulation of complex traits in N. tabacum, laying a foundation for future research on plant functional genomics, crop breeding, and the production of biopharmaceuticals and biofuels.

genetics↗

bioRxiv2023

Improving genome-scale metabolic models of incomplete genomes with deep learning

Deciphering the metabolism of microbial species is crucial for understanding their function within complex ecosystems. Genome-scale metabolic models (GSMMs), which predict metabolic traits based on the enzymes encoded in a genome, are promising tools to study microbial ecosystems when genome sequences can be obtained. However, constructing GSMMs for uncultured bacteria is challenging, as metagenome-assembled genomes are typically incomplete, leading to metabolic reconstructions with numerous gaps. Existing methodologies often fill these gaps with the minimum set of reactions necessary to simulate an objective function such as growth. Here we introduce an artificial intelligence-based alternative: the Deep Neural Network Guided Imputation Of Reactomes (DNNGIOR). The DNNGIOR neural network learns weights for missing reactions in incomplete GSMMs from patterns in the presence and absence of metabolic reactions in genomes spanning the bacterial domain. We identified two important factors contributing to prediction accuracy: (1) the frequency of reaction across all bacteria, and (2) the phylogenetic distance between the query and the genomes in the training dataset. Reactions that occur in > 30% of the training genomes can be accurately predicted (Mean F1 score = 0.85). The weights generated by the DNNGIOR network improved the gap-filling of incomplete GSMMs, when assessed on a large and phylogenetically diverse testing dataset and a small set of high-quality manually curated models. The accuracy of DNNGIOR was on average 14 times greater than the standard unweighted gap-filling for draft reconstructions, and 2-9 times greater for manually curated models. DNNGIOR models could also simulate experimentally measured carbon usage profiles with similar accuracy as CarveMe. DNNGIOR is available at https://github.com/MGXlab/DNNGIOR or as a pip package (https://pypi.org/project/dnngior/).

bioinformatics↗

bioRxiv2023

FLAG: Find, Label Annotate Genomes, a fully automated tool for genome gene structural and functional annotation of highly fragmented non-model species

Recent advances in long-read sequencing technologies and the efforts of projects aimed at increasing the universe of sequenced reference genomes have led to a growth in the number of whole genomes sequenced for non-model organisms. Still, 81% of the over 36,000 unique publicly available eukaryotic genomes in the NCBI database lack gene structure annotations (1). While there are many open-source tools available for each step in the annotation process, many of these tools are designed for chromosomal assemblies with available transcript data from the same or very closely related organisms. Here we present "Find, Label, Annotate Genomes" (FLAG), a fully automated genome annotation workflow. FLAG (i) works on any computing environment, (ii) runs automatically without initial training data, (iii) generates structural and functional annotations, (iv) performs accurately with fragmented genomes, (v) does not require species-specific extrinsic evidence (transcript sequences) and (vi) includes quality control steps to evaluate annotation completeness. We compared the gene annotations generated by FLAG and publicly available annotations from 12 eukaryotic organisms, including 1 plant, 8 living animals, and 1 extinct animal. In fragmented genomes, FLAG annotations provided an average of 18% increase in complete BUSCO scores and 15x lower error rate for the predicted number of protein-coding genes when compared to published BRAKER2 annotations. With FLAG-Refiner further improved results by decreasing the error rate to 17x lower when compared to published BRAKER2 annotations. In high-quality model organisms, FLAG demonstrates comparable results to those of the NCBI EGAP pipeline, underscoring its robustness and accuracy in gene prediction across diverse taxa and assembly qualities.

bioinformatics↗

bioRxiv2023

From populations to pan-genomes: investigating the role of ecology and genomic architecture in maintaining species boundaries in the porcini mushroom, Boletus edulis

Hybridization among diverged species is common across all domains of eukaryotic life, but appears to be particularly common in fungi. However, some fungi exhibit a greater tendency for hybridization than others and it is unclear what mechanisms facilitate or prevent hybridization. Here, we generated 253 whole genome sequences and 22 reference genomes within the globally distributed ectomycorrhizal species complex Boletus edulis Bull. and use a multi-faceted genomic approach to identify patterns of ongoing hybridization and determine whether hybridization is mediated by changes in genome structure, ecology, or both. We found that hybridization is common among species of B. edulis despite 2.1 million years of divergence. However, not all lineages that hybridize today exhibit patterns of introgression, indicating the presence of strong reproductive boundaries among some species. Using a pan-genomic approach we find that genome structural variation is abundant within B. edulis, but that the presence of ongoing introgression does not correlate with genome structural similarity or overall gene content. Instead, we find that the composition of ecologically relevant gene families more accurately predicts the presence of introgression among lineages and B. edulis as a group may specialize on chitin decomposition. Altogether, we show that ecological preferences are likely the primary driver of reproductive barriers in B. edulis.

evolutionary biology↗

bioRxiv2023

Activation of the Maternal Genome Through Asymmetric Distribution of Oocyte-Genome-Associated Histone H3.3

Upon fertilization, the typically silent and epigenetically repressed oocyte genome undergoes activation, yet the precise mechanism remains unclear. The histone variant H3.3 is evenly distributed throughout the oocyte genome, suggesting its involvement in repression. This study reveals that oocyte-genome-associated H3.3 (oH3.3) undergoes asymmetric segregation among four-cell stage blastomeres, persisting in only two blastomeres through the blastocyst stage. These oH3.3-retaining blastomeres maintain a repressive state characterized by high levels of the chromatin marker H3K9me2. Intriguingly, single-cell RNA-seq analysis revealed asymmetric transcriptional activation between paternal and maternal genomes, with the maternal genome being considerably less active. We propose a model wherein oH3.3 and associated oocyte DNA co-segregate during mitosis, allowing two blastomeres to inherit oH3.3 and a strand of oocyte DNA from maternal chromatids. Meanwhile, the other blastomeres acquire newly synthesized DNA associated with the nascent histone H3, which lacks oocyte-specific repressive modifications. Consequently, full maternal genome activation occurs in two of the four-cell stage blastomeres, while the remaining two, which retain oH3.3, remain partially repressed. This study uncovers a previously unrecognized H3.3 mediated mechanism for maternal genome activation.

developmental biology↗

bioRxiv2024

Genomes of Aegilops umbellulata provide new insights into unique structural variations and genetic diversity in the U-genome for wheat improvement

Aegilops spp. serve as an important reservoir for novel sources of resistance or tolerance to biotic and abiotic stresses. To harness this reservoir, we have generated a high-quality chromosome-level genome assembly of an Ae. umbellulata accession PI 554389 using a combination of PacBio HiFi, Oxford nanopore, and chromosome conformation capture (Hi-C) sequencing technologies and resequenced 20 Ae. umbellulata genomes using Illumina sequencing. We assembled a 4.20 Gb genome spanned over seven chromosomes, rich in repetitive elements ([~]84%), achieving a QV of 59.54 with 98.14% completeness. The phylogenetic analysis places the U-genome with D-lineage, but major and distinct rearrangements were revealed in the U-genome. Unique transposon landscape of diploid U-genome and complex chromosomal rearrangements, most prominently in 4U and 6U chromosomes uncovered a distinct evolutionary trajectory of Ae. umbellulata. Additionally, the resequencing of geographically and morphologically diverse Ae. umbellulata accessions revealed three distinctive evolutionary sub-populations. Resequencing also identified six new haplotypes for Lr9, the first leaf rust resistance gene introgressed and cloned from Ae. umbellulata. These genomics resources along with high levels of resistance in the resequenced accessions against five devastating wheat diseases affirmed the genetic potential of Ae. umbellulata for wheat improvement.

plant biology↗

bioRxiv2024

Human immunodeficiency virus-1 induces and targets host genomic R-loops for viral genome integration

Although HIV-1 integration sites favor active transcription units in the human genome, high-resolution analysis of individual HIV-1 integration sites has shown that the virus can integrate into a variety of host genomic locations, including non-genic regions. The invisible infection by HIV-1 integrating into non-genic regions, challenging the traditional understanding of HIV-1 integration site selection, is more problematic because they are selected for preservation in the host genome during prolonged antiretroviral therapies. Here, we showed that HIV-1 integrates its viral genome into the vicinity of R-loops, a genomic structure composed of DNA- RNA hybrids. VSV-G-pseudotyped HIV-1 infection initiates the formation of R-loops in both genic and non-genic regions of the host genome and preferentially integrates into R-loop-rich regions. Using a HeLa cell model that can independently control transcriptional activity and R-loop formation, we demonstrated that the exogenous formation of R-loops directs HIV-1 integration-targeting sites. We also found that HIV-1 integrase proteins physically bind to the host genomic R-loops. These findings provide novel insights into the mechanisms underlying retroviral integration and the new strategies for antiretroviral therapy against HIV-1 latent infection.

molecular biology↗

bioRxiv2024

Are Genomic Language Models All You Need? Exploring Genomic Language Models on Protein Downstream Tasks

Large language models, trained on enormous corpora of biological sequences, are state-of-the-art for downstream genomic and proteomic tasks. Since the genome contains the information to encode all proteins, genomic language models (gLMs) hold the potential to make downstream predictions not only about DNA sequences, but also about proteins. However, the performance of gLMs on protein tasks remains unknown, due to few tasks pairing proteins with the coding DNA sequences (CDS) that can be processed by gLMs. In this work, we curated five such datasets and used them to evaluate the performance of gLMs and proteomic language models (pLMs). We show that gLMs are competitive and even outperform their pLMs counterparts on some tasks. The best performance was achieved using the retrieved CDS compared to sampling strategies. We found that training a joint genomic-proteomic model outperforms each individual approach, showing that they capture different but complementary sequence representations, as we demonstrate through model interpretation of their embeddings. Lastly, we explored different genomic tokenization schemes to improve downstream protein performance. We trained a new Nucleotide Transformer (50M) foundation model with 3mer tokenization that outperforms its 6mer counterpart on protein tasks while maintaining performance on genomics tasks. The application of gLMs to proteomics offers the potential to leverage rich CDS data, and in the spirit of the central dogma, the possibility of a unified and synergistic approach to genomics and proteomics. We make our inference code, model weights and datasets available.

bioinformatics↗

bioRxiv2024

Intra-genomic genes-to-genes correlation enables genome representaion

The bacterial pan-genome consists of core genes shared by all members of a taxonomy and accessory genes found in only a subset. The correlation among genes within the pan-genome could advance our understanding of evolution and tackle medical challenges. Here, we discovered a strong intra-genomic correlation among bacterial pan-genes within each of Escherichia coli, Listeria monocytogenes, Staphylococcus aureus, and Campylobacter jejuni. With a convolutional neural network assisted core genome knock-out simulation, we found that different combinations of fewer than 20 highly variable core genes could recover the sub-species type classified by complete core genome with accuracy >95%. This observation led us to test the genes-to-genes predictability: with more than 52,000 assemblies from each species, combinations of highly variable core genes could predict the sequence variants of other core genes (average accuracy >94%) within the same genome and could also predict sequence variants (average accuracy >91%) as well as the presence (average AUROC >0.91) of some accessory genes. Furthermore, combinations of highly variable core genes could also predict multiple antibiotic resistances (AUROC >0.80) in large published datasets of E. coli, S. aureus, and Mycobacterium tuberculosis. Collectively, we propose that genes within the same genome can strongly correlate with each other. Therefore, the strain phylogeny and the status of other genes could be uniformly represented by combinations of highly variable core genes, which could further represent certain phenotypes including in vitro resistance.

bioinformatics↗

bioRxiv2025

A k-mer-based maximum likelihood method for estimating distances of reads to genomes enables genome-wide phylogenetic placement.

Comparing each sequencing read in a sample to large databases of known genomes has become a fundamental tool with wide-ranging applications, including metagenomics. These comparisons can be based on read-to-genome alignment, which is relatively slow, especially if done with the high sensitivity needed to characterize queries without a close representation in the reference dataset. A more scalable alternative is assigning taxonomic labels to reads using signatures such as k-mer presence/absence. A third approach is placing reads on a reference phylogeny, which can provide a far more detailed view of the read than a single label. How-ever, phylogenetic placement is currently only possible at scale for marker genes, constituting a small fraction of the genome. No current method is able to place all reads originating from anywhere in the genome on an ultra-large reference phylogeny. In this paper, we introduce krepp, an alignment-free k-mer-based method that enables placing reads from anywhere on the genome on an ultra-large reference phylogeny by first computing a distance from each read to every reference genome. To compute these distances and placements, krepp uses a host of algorithmic techniques, including locality-sensitive hashing to allow inexact k-mer matches, k-mer coloring graphs to map k-mers to reference genomes, maximum likelihood distance estimation, and likelihood ratio test for placement. Our experiments show that krepp is extremely scalable, improving on alignment by up to roughly 10x, computes very accurate distances that approximate those using alignments, and produces highly accurate placements. When used in the metagenomics context, the precise phylogenetic identifications provided by krepp improve our ability to compare and differentiate samples from different environments.

bioinformatics↗