TERL: classification of transposable elements by convolutional neural networks

Genomes of the Orestias pupfish from the Andean Altiplano shed light on their evolutionary history and phylogenetic relationships within Cyprinodontiformes

BACKGROUND

To unravel the evolutionary history of a complex group, a comprehensive reconstruction of its phylogenetic relationships is crucial. This requires meticulous taxon sampling and careful consideration of multiple characters to ensure a complete and accurate reconstruction. The phylogenetic position of the Orestias genus has been estimated partly on unavailable or incomplete information. As a consequence, it was assigned to the family Cyprindontidae, relating this Andean fish to other geographically distant genera distributed in the Mediterranean, Middle East and North and Central America. In this study, using complete genome sequencing, we aim to clarify the phylogenetic position of Orestias within the Cyprinodontiformes order.

RESULTS

We sequenced the genome of three Orestias species from the Andean Altiplano. Our analysis revealed that the small genome size in this genus (~ 0.7 Gb) was caused by a contraction in transposable element (TE) content, particularly in DNA elements and short interspersed nuclear elements (SINEs). Using predicted gene sequences, we generated a phylogenetic tree of Cyprinodontiformes using 902 orthologs extracted from all 32 available genomes as well as three outgroup species. We complemented this analysis with a phylogenetic reconstruction and time calibration considering 12 molecular markers (eight nuclear and four mitochondrial genes) and a stratified taxon sampling to consider 198 species of nearly all families and genera of this order. Overall, our results show that phylogenetic closeness is directly related to geographical distance. Importantly, we found that Orestias is not part of the Cyprinodontidae family, and that it is more closely related to the South American fish fauna, being the Fluviphylacidae the closest sister group.

CONCLUSIONS

The evolutionary history of the Orestias genus is linked to the South American ichthyofauna and it should no longer be considered a member of the Cyprinodontidae family. Instead, we submit that Orestias belongs to the Orestiidae family, as suggested by Freyhof et al. (2017), and that it is the sister group of the Fluviphylacidae family, distributed in the Amazonian and Orinoco basins. These two groups likely diverged during the Late Eocene concomitant with hydrogeological changes in the South American landscape.

The genomes of seven economic Caesalpinioideae trees provide insights into polyploidization history and secondary metabolite biosynthesis

The Caesalpinioideae subfamily contains many well-known trees that are important for economic sustainability and human health, but a lack of genomic resources has hindered their breeding and utilization. Here, we present chromosome-level reference genomes for the two food and industrial trees Gleditsia sinensis (921 Mb) and Biancaea sappan (872 Mb), the three shade and ornamental trees Albizia julibrissin (705 Mb), Delonix regia (580 Mb), and Acacia confusa (566 Mb), and the two pioneer and hedgerow trees Leucaena leucocephala (1338 Mb) and Mimosa bimucronata (641 Mb). Phylogenetic inference shows that the mimosoid clade has a much higher evolutionary rate than the other clades of Caesalpinioideae. Macrosynteny comparison suggests that the fusion and breakage of an unstable chromosome are responsible for the difference in basic chromosome number (13 or 14) for Caesalpinioideae. After an ancient whole-genome duplication (WGD) shared by all Caesalpinioideae species (CWGD, ∼72.0 million years ago [MYA]), there were two recent successive WGD events, LWGD-1 (16.2-19.5 MYA) and LWGD-2 (7.1-9.5 MYA), in L. leucocephala. Thereafter, ∼40% gene loss and genome-size contraction have occurred during the diploidization process in L. leucocephala. To investigate secondary metabolites, we identified all gene copies involved in mimosine metabolism in these species and found that the abundance of mimosine biosynthesis genes in L. leucocephala largely explains its high mimosine production. We also identified the set of all potential genes involved in triterpenoid saponin biosynthesis in G. sinensis, which is more complete than that based on previous transcriptome-derived unigenes. Our results and genomic resources will facilitate biological studies of Caesalpinioideae and promote the utilization of valuable secondary metabolites.

From tradition to innovation: conventional and deep learning frameworks in genome annotation

Following the milestone success of the Human Genome Project, the 'Encyclopedia of DNA Elements (ENCODE)' initiative was launched in 2003 to unearth information about the numerous functional elements within the genome. This endeavor coincided with the emergence of numerous novel technologies, accompanied by the provision of vast amounts of whole-genome sequences, high-throughput data such as ChIP-Seq and RNA-Seq. Extracting biologically meaningful information from this massive dataset has become a critical aspect of many recent studies, particularly in annotating and predicting the functions of unknown genes. The core idea behind genome annotation is to identify genes and various functional elements within the genome sequence and infer their biological functions. Traditional wet-lab experimental methods still rely on extensive efforts for functional verification. However, early bioinformatics algorithms and software primarily employed shallow learning techniques; thus, the ability to characterize data and features learning was limited. With the widespread adoption of RNA-Seq technology, scientists from the biological community began to harness the potential of machine learning and deep learning approaches for gene structure prediction and functional annotation. In this context, we reviewed both conventional methods and contemporary deep learning frameworks, and highlighted novel perspectives on the challenges arising during annotation underscoring the dynamic nature of this evolving scientific landscape.

Haplotype-resolved chromosome-level genome of hexaploid Jerusalem artichoke provides insights into its origin, evolution, and inulin metabolism

Jerusalem artichoke (Helianthus tuberosus) is a global multifunctional crop. It has wide applications in the food, health, feed, and biofuel industries and in ecological protection; it also serves as a germplasm pool for breeding of the global oil crop common sunflower (Helianthus annuus). However, biological studies of Jerusalem artichoke have been hindered by a lack of genome sequences, and its high polyploidy and large genome size have posed challenges to genome assembly. Here, we report a 21-Gb chromosome-level assembly of the hexaploid Jerusalem artichoke genome, which comprises 17 homologous groups, each with 6 pseudochromosomes. We found multiple large-scale chromosome rearrangements between Jerusalem artichoke and common sunflower, and our results show that the hexaploid genome of Jerusalem artichoke was formed by a hybridization event between a tetraploid and a diploid Helianthus species, followed by chromosome doubling of the hybrid, which occurred approximately 2 million years ago. Moreover, we identified more copies of actively expressed genes involved in inulin metabolism and showed that these genes may still be undergoing loss of function or sub- or neofunctionalization. These genomic resources will promote further biological studies, breeding improvement, and industrial utilization of Helianthus crops.

Genomic object detection: An improved approach for transposable elements detection and classification using convolutional neural networks

Analysis of eukaryotic genomes requires the detection and classification of transposable elements (TEs), a crucial but complex and time-consuming task. To improve the performance of tools that accomplish these tasks, Machine Learning approaches (ML) that leverage computer resources, such as GPUs (Graphical Processing Unit) and multiple CPU (Central Processing Unit) cores, have been adopted. However, until now, the use of ML techniques has mostly been limited to classification of TEs. Herein, a detection-classification strategy (named YORO) based on convolutional neural networks is adapted from computer vision (YOLO) to genomics. This approach enables the detection of genomic objects through the prediction of the position, length, and classification in large DNA sequences such as fully sequenced genomes. As a proof of concept, the internal protein-coding domains of LTR-retrotransposons are used to train the proposed neural network. Precision, recall, accuracy, F1-score, execution times and time ratios, as well as several graphical representations were used as metrics to measure performance. These promising results open the door for a new generation of Deep Learning tools for genomics. YORO architecture is available at https://github.com/simonorozcoarias/YORO.

Repetitive DNA sequence detection and its role in the human genome

Repetitive DNA sequences playing critical roles in driving evolution, inducing variation, and regulating gene expression. In this review, we summarized the definition, arrangement, and structural characteristics of repeats. Besides, we introduced diverse biological functions of repeats and reviewed existing methods for automatic repeat detection, classification, and masking. Finally, we analyzed the type, structure, and regulation of repeats in the human genome and their role in the induction of complex diseases. We believe that this review will facilitate a comprehensive understanding of repeats and provide guidance for repeat annotation and in-depth exploration of its association with human diseases.

Chromosome-scale Amaranthus tricolor genome provides insights into the evolution of the genus Amaranthus and the mechanism of betalain biosynthesis

Amaranthus tricolor is a vegetable and ornamental amaranth, with high lysine, dietary fibre and squalene content. The red cultivar of A. tricolor possesses a high concentration of betalains, which has been used as natural food colorants. Here, we constructed the genome of A. tricolor, the first reference genome for the subgenus Albersia, combining PacBio HiFi, Nanopore ultra-long and Hi-C data. The contig N50 size was 906 kb, and 99.58% of contig sequence was anchored to the 17 chromosomes, totalling 520 Mb. We annotated 27,813 protein-coding genes with an average 1.3 kb coding sequence and 5.3 exons. We inferred that A. tricolor underwent a whole-genome duplication (WGD) and that the WGD shared by amaranths occurred in the last common ancestor of subfamily Amaranthoideae. Moreover, we comprehensively identified candidate genes in betalain biosynthesis pathway. Among them, DODAα1 and CYP76ADα1, located in one topologically associated domain (TAD) of an active (A) compartment on chromosome 16, were more highly expressed in red leaves than in green leaves, and DODAα1 might be the rate-limiting enzyme gene in betalains biosynthesis. This study presents new genome resources and enriches our understanding of amaranth evolution, betalains production, facilitating molecular breeding improvements and the understanding of C4 plants evolution.

Chromosome-level genome of a leaf vegetable Glebionis coronaria provides insights into the biosynthesis of monoterpenoids contributing to its special aroma

Glebionis coronaria is a popular vegetable with special aroma and a medical plant in East Asia and Mediterranean, but its biological studies and breeding have been hindered by the lack of reference genome. Here, we present a chromosome-level reference genome of G. coronaria, with assembled genome size of 6.8 Gb, which is the largest among all the published genomes of diploid Asteraceae species. The large genome size of G. coronaria is mainly caused by the recent widespread explosions of long-terminal-repeat retrotransposons. Analyses of macro-synteny and synonymous mutation rate distribution indicate that the G. coronaria genome experienced a whole-genome triplication at 40-45 million years ago, shared with all Asteraceae species. In subtribe Artemisiinae, Glebionis arose before the divergence of Chrysanthemum from Artemisia, and Glebionis species evolved much faster than Chrysanthemum and Artemisia species. In G. coronaria, the synthesis genes of monoterpenoids 8-oxocitronellyl enol and isopiperitenone were expanded, and the higher expressions of these expanded genes in leaves and stems may contribute to its special aroma. The G. coronaria genomic resources will promote the evolution studies of Asteraceae, the metabolism mechanism studies of bioactive compounds, and the breeding improvement of agronomic traits in G. coronaria.

The genomes of chicory, endive, great burdock and yacon provide insights into Asteraceae palaeo-polyploidization history and plant inulin production

Inulin is an important reserve polysaccharide in Asteraceae plants, and is also widely used as a sweetener, a source of dietary fibre and prebiotic. Nevertheless, a lack of genomic resources for inulin-producing plants has hindered extensive studies on inulin metabolism and regulation. Here, we present chromosome-level reference genomes for four inulin-producing plants: chicory (Cichorium intybus), endive (Cichorium endivia), great burdock (Arctium lappa) and yacon (Smallanthus sonchifolius), with assembled genome sizes of 1.28, 0.89, 1.73 and 2.72 Gb, respectively. We found that the chicory, endive and great burdock genomes were shaped by whole genome triplication (WGT-1), and the yacon genome was shaped by WGT-1 and two subsequent whole genome duplications (WGD-2 and WGD-3). A yacon unique whole genome duplication (WGD-3) occurred 5.6-5.8 million years ago. Our results also showed the genome size difference between chicory and endive is largely due to LTR retrotransposons, and rejected a previous hypothesis that chicory is an ancestor of endive. Furthermore, we identified fructan-active-enzyme and transcription-factor genes, and found there is one copy in chicory, endive and great burdock but two copies in yacon for most of these genes, except for the 1-FEH II gene which is significantly expanded in chicory. Interestingly, inulin synthesis genes 1-SST and 1-FFT are located close to each other, as are the degradation genes 1-FEH I and 1-FEH II. Finally, we predicted protein structures for 1-FFT genes to explore the mechanism determining inulin chain length.

A chromosome-level reference genome of a Convolvulaceae species Ipomoea cairica

Ipomoea cairica is a perennial creeper that has been widely introduced as a garden ornamental across tropical, subtropical, and temperate regions. Because it grows extremely fast and spreads easily, it has been listed as an invasive species in many countries. Here, we constructed the chromosome-level reference genome of Ipomoea cairica by Pacific Biosciences HiFi and Hi-C sequencing, with the assembly size of 733.0 Mb, the contig N50 of 43.8 Mb, the scaffold N50 of 45.7 Mb, and the Benchmarking Universal Single-Copy Orthologs complete rate of 98.0%. Hi-C scaffolding assigned 97.9% of the contigs to 15 pseudo-chromosomes. Telomeric repeat analysis reveals that 7 of the 15 pseudo-chromosomes are gapless and telomere to telomere. The transposable element content of Ipomoea cairica is 73.4%, obviously higher than that of other Ipomoea species. A total of 38,115 protein-coding genes were predicted, with the Benchmarking Universal Single-Copy Orthologs complete rate of 98.5%, comparable to that of the genome assembly, and 92.6% of genes were functional annotated. In addition, we identified 3,039 tRNA genes and 2,403 rRNA genes in the assembled genome. Phylogenetic analysis showed that Ipomoea cairica formed a clade with Ipomoea aquatica, and they diverged from each other 8.1 million years ago. Through comparative genome analysis, we reconfirmed that a whole genome triplication event occurred specific to Convolvulaceae family and in the ancestor of the genus Ipomoea and Cuscuta. This high-quality reference genome of Ipomoea cairica will greatly facilitate the studies on the molecular mechanisms of its rapid growth and invasiveness.

Automatic curation of LTR retrotransposon libraries from plant genomes through machine learning

Transposable elements are mobile sequences that can move and insert themselves into chromosomes, activating under internal or external stimuli, giving the organism the ability to adapt to the environment. Annotating transposable elements in genomic data is currently considered a crucial task to understand key aspects of organisms such as phenotype variability, species evolution, and genome size, among others. Because of the way they replicate, LTR retrotransposons are the most common transposable elements in plants, accounting in some cases for up to 80% of all DNA information. To annotate these elements, a reference library is usually created, a curation process is performed, eliminating TE fragments and false positives and then annotated in the genome using the homology method. However, the curation process can take weeks, requires extensive manual work and the execution of multiple time-consuming bioinformatics software. Here, we propose a machine learning-based approach to perform this process automatically on plant genomes, obtaining up to 91.18% F1-score. This approach was tested with four plant species, obtaining up to 93.6% F1-score (Oryza granulata) in only 22.61 s, where bioinformatics methods took approximately 6 h. This acceleration demonstrates that the ML-based approach is efficient and could be used in massive sequencing projects.

TransposonUltimate: software for transposon classification, annotation and detection

Most genomes harbor a large number of transposons, and they play an important role in evolution and gene regulation. They are also of interest to clinicians as they are involved in several diseases, including cancer and neurodegeneration. Although several methods for transposon identification are available, they are often highly specialised towards specific tasks or classes of transposons, and they lack common standards such as a unified taxonomy scheme and output file format. We present TransposonUltimate, a powerful bundle of three modules for transposon classification, annotation, and detection of transposition events. TransposonUltimate comes as a Conda package under the GPL-3.0 licence, is well documented and it is easy to install through https://github.com/DerKevinRiehl/TransposonUltimate. We benchmark the classification module on the large TransposonDB covering 891,051 sequences to demonstrate that it outperforms the currently best existing solutions. The annotation and detection modules combine sixteen existing softwares, and we illustrate its use by annotating Caenorhabditis elegans, Rhizophagus irregularis and Oryza sativa subs. japonica genomes. Finally, we use the detection module to discover 29 554 transposition events in the genomes of 20 wild type strains of C. elegans. Databases, assemblies, annotations and further findings can be downloaded from (https://doi.org/10.5281/zenodo.5518085).

Finding and Characterizing Repeats in Plant Genomes

Plant genomes contain a particularly high proportion of repeated structures of various types. This chapter proposes a guided tour of the available software that can help biologists to scan automatically for these repeats in sequence data or check hypothetical models intended to characterize their structures. Since transposable elements (TEs) are a major source of repeats in plants, many methods have been used or developed for this broad class of sequences. They are representative of the range of tools available for other classes of repeats and we have provided two sections on this topic (for the analysis of genomes or directly of sequenced reads), as well as a selection of the main existing software. It may be hard to keep up with the profusion of proposals in this dynamic field and the rest of the chapter is devoted to the foundations of an efficient search for repeats and more complex patterns. We first introduce the key concepts of the art of indexing and mapping or querying sequences. We end the chapter with the more prospective issue of building models of repeat families. We present the Machine Learning approach first, seeking to build predictors automatically for some families of ET, from a set of sequences known to belong to this family. A second approach, the linguistic (or syntactic) approach, allows biologists to describe themselves and check the validity of models of their favorite repeat family.

Computational Analysis of Transposable Elements and CircRNAs in Plants

This chapter provides two main contributions: (1) a description of computational tools and databases used to identify and analyze transposable elements (TEs) and circRNAs in plants; and (2) data analysis on public TE and circRNA data. Our goal is to highlight the primary information available in the literature on circular noncoding RNAs and transposable elements in plants. The exploratory analysis performed on publicly available circRNA and TEs data help discuss four sequence features. Finally, we investigate the association on circRNAs:TE in plants in the model organism Arabidopsis thaliana.

Bioinformatics Approaches for Determining the Functional Impact of Repetitive Elements on Non-coding RNAs

With a large number of annotated non-coding RNAs (ncRNAs), repetitive sequences are found to constitute functional components (termed as repetitive elements) in ncRNAs that perform specific biological functions. Bioinformatics analysis is a powerful tool for improving our understanding of the role of repetitive elements in ncRNAs. This chapter summarizes recent findings that reveal the role of repetitive elements in ncRNAs. Furthermore, relevant bioinformatics approaches are systematically reviewed, which promises to provide valuable resources for studying the functional impact of repetitive elements on ncRNAs.

K-mer counting and curated libraries drive efficient annotation of repeats in plant genomes

The annotation of repetitive sequences within plant genomes can help in the interpretation of observed phenotypes. Moreover, repeat masking is required for tasks such as whole-genome alignment, promoter analysis, or pangenome exploration. Although homology-based annotation methods are computationally expensive, k-mer strategies for masking are orders of magnitude faster. Here, we benchmarked a two-step approach, where repeats were first called by k-mer counting and then annotated by comparison to curated libraries. This hybrid protocol was tested on 20 plant genomes from Ensembl, with the k-mer-based Repeat Detector (Red) and two repeat libraries (REdat, last updated in 2013, and nrTEplants, curated for this work). Custom libraries produced by RepeatModeler were also tested. We obtained repeated genome fractions that matched those reported in the literature but with shorter repeated elements than those produced directly by sequence homology. Inspection of the masked regions that overlapped genes revealed no preference for specific protein domains. Most Red-masked sequences could be successfully classified by sequence similarity, with the complete protocol taking less than 2 h on a desktop Linux box. A guide to curating your own repeat libraries and the scripts for masking and annotating plant genomes can be obtained at https://github.com/Ensembl/plant-scripts.

K-mer-based machine learning method to classify LTR-retrotransposons in plant genomes

Every day more plant genomes are available in public databases and additional massive sequencing projects (i.e., that aim to sequence thousands of individuals) are formulated and released. Nevertheless, there are not enough automatic tools to analyze this large amount of genomic information. LTR retrotransposons are the most frequent repetitive sequences in plant genomes; however, their detection and classification are commonly performed using semi-automatic and time-consuming programs. Despite the availability of several bioinformatic tools that follow different approaches to detect and classify them, none of these tools can individually obtain accurate results. Here, we used Machine Learning algorithms based on k-mer counts to classify LTR retrotransposons from other genomic sequences and into lineages/families with an F1-Score of 95%, contributing to develop a free-alignment and automatic method to analyze these sequences.

InpactorDB: A Classified Lineage-Level Plant LTR Retrotransposon Reference Library for Free-Alignment Methods Based on Machine Learning

Long terminal repeat (LTR) retrotransposons are mobile elements that constitute the major fraction of most plant genomes. The identification and annotation of these elements via bioinformatics approaches represent a major challenge in the era of massive plant genome sequencing. In addition to their involvement in genome size variation, LTR retrotransposons are also associated with the function and structure of different chromosomal regions and can alter the function of coding regions, among others. Several sequence databases of plant LTR retrotransposons are available for public access, such as PGSB and RepetDB, or restricted access such as Repbase. Although these databases are useful to identify LTR-RTs in new genomes by similarity, the elements of these databases are not fully classified to the lineage (also called family) level. Here, we present InpactorDB, a semi-curated dataset composed of 130,439 elements from 195 plant genomes (belonging to 108 plant species) classified to the lineage level. This dataset has been used to train two deep neural networks (i.e., one fully connected and one convolutional) for the rapid classification of these elements. In lineage-level classification approaches, we obtain up to 98% performance, indicated by the F1-score, precision and recall scores.