Thursday, November 19, 2026
Science
No Result
View All Result
  • Login
  • HOME
  • SCIENCE NEWS
  • CONTACT US
  • HOME
  • SCIENCE NEWS
  • CONTACT US
No Result
View All Result
Scienmag
No Result
View All Result
Home Science News

TE Bench Standardizes Performance Testing for Transposable Element Annotation Workflows

November 19, 2026
in Science News
Juliet Wilcox
By Juliet Wilcox Scienmag Editorial Profile - Human Genetics
Reading Time: 6 mins read
0
TE Bench Standardizes Performance Testing for Transposable Element Annotation Workflows

TE Bench Standardizes Performance Testing for Transposable Element Annotation Workflows

65
SHARES
587
VIEWS
Share on FacebookShare on Twitter
ADVERTISEMENT

Computational biologists have introduced TE_Bench, an automated, open-source benchmarking workflow built on the Snakemake management system to systematically evaluate and compare software pipelines designed for transposable element discovery in genomic sequences. Transposable elements represent ubiquitous and dynamic sequence components across diverse genomes, acting as powerful agents of structural diversification, regulatory innovation, and genomic disruption while also contributing to pathological processes such as inflammation and oncogenesis. Despite their fundamental importance across eukaryotic genomics, annotating these elements remains notoriously difficult because transposable elements differ substantially in evolutionary age, structural fragmentation, sequence divergence, subfamily architecture, and nested configurations where younger elements physically insert into older host sequences. While numerous automated pipelines have been released to identify repetitive elements without relying entirely on labor-intensive manual curation, differences in their underlying algorithms are frequently opaque and yield highly divergent results, creating an urgent demand for standardized, reproducible evaluation frameworks.

To overcome the lack of unified assessment standards across computational genomics, researchers Hannah P. Kania, Sierra A. Seifert, and Anne D. Yoder structured TE_Bench into a modular architecture spanning three sequential execution stages that handle sequence simulation, reference annotation generation, and quantitative benchmarking. Operating under Snakemake version 9.11.4 or higher within high-performance computing environments managed by SLURM, the workflow allows researchers to generate customizable synthetic DNA sequences using an adapted version of the GARLIC simulation software. Because rigorously curated, complete reference annotations do not exist for the vast majority of sequenced organisms, synthetic sequence generation provides an objective, mathematically defined ground truth. Users can supply model genomic files downloaded from public repositories such as the UCSC Genome Browser alongside transposable element consensus libraries sourced from either Dfam or Repbase, enabling GARLIC scripts to capture empirical distributions of tetramers, repetitive insertions, and simple sequence repeats across defined ten-kilobase GC-content bins.

The sequence generation phase of the workflow relies on modified Perl scripts from the GARLIC package to simulate non-functional yet biologically realistic DNA sequences embedded with artificially mutated and nested transposable elements. The rule execution plan automated within Simulation_Snakefile reads user-specified configuration files or command-line parameters, accepting inputs that designate sequence length in nucleotides, total simulated sequence counts, and database choices between open-source Dfam libraries and subscription-gated Repbase libraries. Following sequence synthesis, internal workflow scripts such as seqkit.sh and split_inserts.py separate multi-FASTA output files and associated insertion coordinate logs. These intermediate files subsequently feed into AnnotationGen_Snakefile, which automates the construction of standardized reference annotation files. Operating through custom Bash and Awk scripts alongside NCBI Blast+, this second stage resolves internal insertion junctions and nested boundaries that base GARLIC outputs leave unannotated, ultimately converting structural outputs via R scripts into uniform eight-column comma-separated tabular files.

Once reference coordinate files are established, the third phase of the workflow, managed by Benchmark_Snakefile, performs extensive comparative evaluations divided into comprehensive and selective operational modes. In comprehensive benchmarking, the software evaluates all annotated transposable element families across the entire sequence length, whereas selective benchmarking isolates user-designated subsets, such as specific long terminal repeat retrotransposons, using attribute-filtering scripts. To support broad community integration, TE_Bench incorporates conversion scripts that ingest native output formats from widely used de novo discovery pipelines, specifically transforming annotation files from EDTA, RepeatModeler2, and Earl Grey into the required eight-column tabular structure. The software then computes nucleotide-level confusion matrices across the underlying sequences, determining counts for true positives, true negatives, false positives, and false negatives to derive standard statistical metrics including sensitivity, specificity, accuracy, precision, false discovery rate, F1 score, and Matthew’s correlation coefficient.

To validate the utility of TE_Bench, the authors conducted an extensive empirical case study testing three prominent de novo annotation tools—EDTA version 2.2.0, RepeatModeler2 version 2.0.5, and Earl Grey version 4.4.0—across five simulated one-hundred-megabase sequences modeled on the genomes of Arabidopsis thaliana, Drosophila melanogaster, Homo sapiens, Myotis lucifugus, and Saccharomyces cerevisiae. Across the broader comparative set, RepeatModeler2 and Earl Grey demonstrated comparable overall efficacy with mean Matthew’s correlation coefficients of 0.85 and 0.83 respectively, consistently surpassing the default performance of EDTA, which recorded an average coefficient of 0.64. Absolute pipeline performance fluctuated across genomic models; for instance, all pipelines achieved their highest annotation accuracy on the S. cerevisiae sequence, which harbors a simpler landscape restricted to Y-Prime and long terminal repeat elements, whereas sensitivity dropped notably in the human sequence due to the abundance of truncated, non-LTR retrotransposons such as LINE and SINE insertions.

Beyond high-level statistical indices, TE_Bench generates specialized diagnostic visualizations that expose class-specific strengths and algorithmic deficiencies within candidate pipelines. In an analysis of the simulated Myotis lucifugus genome, class-stratified coverage pie charts revealed that while Earl Grey leveraged its RepeatModeler2-based discovery engine to improve boundary detection for DNA transposons and LINE elements, it unexpectedly failed to annotate more than fifty percent of the LINE-dependent retrotransposons correctly recovered by RepeatModeler2. A similar loss of detection sensitivity occurred in the simulated Arabidopsis thaliana genome but was absent in the human sequence. Detailed inspection revealed that Earl Grey version 4.4.0 contained a hardcoded command-line parameter within its RepeatMasker implementation that deliberately suppressed annotations of small RNA genes. While human Alu elements are derived from 7SL RNA and remained unaffected, the predominant SINE elements in bats and plants—specifically Ves and RatheE1—are tRNA-derived and were filtered out. When the investigators disabled this masking flag, Earl Grey’s recovery of LINE-dependent elements in the bat model jumped from 32.6 percent to 87.1 percent.

The benchmarking suite also incorporates specialized analytical routines to assess how effectively annotation programs resolve nested transposable elements, where one mobile sequence has inserted directly into the body of an existing host element. TE_Bench separates reference elements into nested sequences, host sequences, and unnested control sequences, plotting the distribution of recovered coverage across defined percentage bins. Evaluating the bat simulation demonstrated a striking bimodal distribution for nested elements: candidate pipelines either identified them with high fidelity, achieving greater than eighty percent sequence coverage, or missed them almost entirely below twenty percent coverage. Conversely, coverage across host elements exhibited a dispersed, flat distribution across intermediate coverage intervals without a distinct mode. This pattern indicated that automated tools frequently struggle to resolve the disrupted boundaries of older host elements, which have accumulated substantial post-insertion divergence, underscoring the risk of structural misannotation in eukaryotic genomes rich in terminal inverted repeat elements.

Sequence divergence from consensus sequences represents another critical biological dimension assessed by TE_Bench’s regression modules. By leveraging simulated insertion histories generated by GARLIC, the workflow plots nucleotide identity between reference elements and their ancestral consensus against the proportion of sequence covered by each test pipeline. In the bat genomic simulation, regression analyses revealed a positive correlation (Pearson’s r = 0.43) between the sequence identity of DNA transposons and Earl Grey’s annotation coverage, demonstrating that mobile elements with higher sequence conservation relative to consensus sequences are recovered far more reliably. Because EDTA, RepeatModeler2, and Earl Grey all rely on RepeatScout and Recon modules for initial de novo non-LTR repeat detection, the pipelines exhibited similar sensitivity thresholds across varying degrees of mutation. Nevertheless, these divergence plots provide an indispensable analytical diagnostic for benchmarking emerging discovery algorithms against heavily decayed, ancient genomic repeats.

To evaluate annotation over-segmentation and taxonomic confusion, TE_Bench generates quantitative redundancy summaries and cross-class overlap heatmaps. These diagnostics calculate normalized overlap ratios between reference classifications and test predictions, as well as pairwise overlaps within the test annotation itself. In the Myotis lucifugus trial, the resulting matrices exposed substantial off-diagonal misclassification by EDTA, which exhibited an overlap score of 0.61 between true reference DNA transposons and predicted long terminal repeat annotations. Furthermore, the test-versus-test heatmap revealed within-annotation overlap values exceeding 1.0 for EDTA across DNA, LTR, and LINE categories. Such values indicate that fragmented, redundant, and overlapping element predictions repeatedly covered identical genomic base pairs, thereby falsely inflating base-pair sensitivity metrics while severely degrading locus-level architectural precision and boundary delineation.

Demonstrating its versatility on biological sequencing data, TE_Bench was applied to evaluate how primary sequencing technologies and bioinformatic read alignment strategies affect downstream transposable element discovery in real genomes. Utilizing Illumina 250-base-pair paired-end reads and PacBio HiFi long reads derived from Drosophila melanogaster ISO1 stock, researchers aligned sequences against release 6 of the reference genome assembly using BWA, minimap2, and pbmm2 under default configurations. After generating consensus sequences with samtools consensus and annotating them across all three pipelines against the published reference annotation, the analysis revealed that consensus assemblies constructed from long reads aligned with BWA yielded the highest Matthew’s correlation coefficients across all three tools. Intriguingly, consensus assemblies built from PacBio reads mapped with minimap2 or pbmm2 produced lower correlation scores than assemblies generated from short Illumina reads aligned with BWA, corroborating prior observations that default minimap2 parameters occasionally misalign reads spanning highly repetitive loci.

In evaluating workflow constraints and operational boundaries, the authors observed that TE_Bench currently executes benchmarking operations primarily at the base-pair level, meaning that locus-level assembly completeness and family-resolved clustering require complementary qualitative inspection. Additionally, because the GARLIC sequence simulation engine operates non-deterministically without a fixed pseudorandom seed, the researchers analyzed nineteen independently simulated 100-megabase Drosophila sequences across two distinct model builds. The resulting 95 percent confidence intervals demonstrated minimal variance across all performance indices, with the widest observed interval width reaching only 0.030188 for Earl Grey’s sensitivity, confirming that simulation variability does not compromise pipeline rankings. Released under open-source MIT and GNU GPL-3.0 licenses, TE_Bench establishes a rigorous, community-accessible foundation for vetting transposable element discovery software across diverse simulated landscapes and empirical sequencing platforms.

Subject of Research: Benchmarking workflow for computational transposable element annotation pipelines

Article Title: TE_Bench: a foundational benchmarking workflow for transposable element annotation pipelines

Article References: Kania, H. P., Seifert, S. A., & Yoder, A. D. (2026). TE_Bench: a foundational benchmarking workflow for transposable element annotation pipelines. Mobile DNA, 17(1), Article 27. https://doi.org/10.1186/s13100-026-00405-z

Image Credits: AI Generated

DOI: 10.1186/s13100-026-00405-z

Keywords: TE_Bench, transposable elements, genome annotation, bioinformatics pipeline, Snakemake, RepeatModeler2, Earl Grey, EDTA, GARLIC simulator, genomic repeats, Matthew's correlation coefficient, sequence divergence, nested transposable elements, read mapping, PacBio HiFi, BWA aligner, Drosophila melanogaster

Cite Scienmag News

Juliet Wilcox. (November 19, 2026). TE Bench Standardizes Performance Testing for Transposable Element Annotation Workflows. Scienmag. https://scienmag.com/te-bench-standardizes-performance-testing-for-transposable-element-annotation-workflows/

Juliet Wilcox. "TE Bench Standardizes Performance Testing for Transposable Element Annotation Workflows." Scienmag, 19 November 2026, https://scienmag.com/te-bench-standardizes-performance-testing-for-transposable-element-annotation-workflows/. Accessed 19 November 2026.

Juliet Wilcox. "TE Bench Standardizes Performance Testing for Transposable Element Annotation Workflows." Scienmag. November 19, 2026. https://scienmag.com/te-bench-standardizes-performance-testing-for-transposable-element-annotation-workflows/

Tags: bioinformatics pipelineBWA alignerDrosophila melanogasterEarl GreyEDTAGARLIC simulatorgenome annotationgenomic repeatsMatthews Correlation Coefficientnested transposable elementsPacBio HiFiread mappingRepeatModeler2sequence divergenceSnakemakeTE_Benchtransposable elements
Share26Tweet16
Previous Post

Periodontitis Reveals Immune Heterogeneity Through Fusobacterium-Reactive Antibody Profiles

Next Post

Tailored Passivation Molecules Boost Perovskite Solar Cell Efficiency Above Twenty-Five Percent

Related Posts

Chronic Breathlessness Linked to Increased Financial Stress in Australian Survey
Medicine

Chronic Breathlessness Linked to Increased Financial Stress in Australian Survey

November 19, 2026
Deep Learning Electrocardiograms Underestimate Mortality Risk Under Clinical Case-Mix Shift
Medicine

Deep Learning Electrocardiograms Underestimate Mortality Risk Under Clinical Case-Mix Shift

November 19, 2026
Female Traffic Deaths in Iran Linked to 121,929 Potential Foregone Births
Medicine

Female Traffic Deaths in Iran Linked to 121,929 Potential Foregone Births

November 19, 2026
Hybrid Deep Learning Model Detects Narcoleptic Sleep Irregularities With High Accuracy
Technology and Engineering

Hybrid Deep Learning Model Detects Narcoleptic Sleep Irregularities With High Accuracy

November 19, 2026
AI Capacity Shows Conditional Link to Environmental Sustainability in Qatar Study
Earth Science

AI Capacity Shows Conditional Link to Environmental Sustainability in Qatar Study

November 19, 2026
Retinal Pigment Epithelial Receptor Activation Stimulates Mouse Müller Glia Reprogramming
Medicine

Retinal Pigment Epithelial Receptor Activation Stimulates Mouse Müller Glia Reprogramming

November 19, 2026
Next Post
Tailored Passivation Molecules Boost Perovskite Solar Cell Efficiency Above Twenty-Five Percent

Tailored Passivation Molecules Boost Perovskite Solar Cell Efficiency Above Twenty-Five Percent

  • Mothers who receive childcare support from maternal grandparents show more optimized

    Mothers who receive childcare support from maternal grandparents show more parental warmth, finds NTU Singapore study

    27656 shares
    Share 11059 Tweet 6912
  • University of Seville Breaks 120-Year-Old Mystery, Revises a Key Einstein Concept

    1061 shares
    Share 424 Tweet 265
  • Bee body mass, pathogens and local climate influence heat tolerance

    682 shares
    Share 273 Tweet 171
  • Researchers record first-ever images and data of a shark experiencing a boat strike

    546 shares
    Share 218 Tweet 137
  • Groundbreaking Clinical Trial Reveals Lubiprostone Enhances Kidney Function

    531 shares
    Share 212 Tweet 133
Science

Embark on a thrilling journey of discovery with Scienmag.com—your ultimate source for cutting-edge breakthroughs. Immerse yourself in a world where curiosity knows no limits and tomorrow’s possibilities become today’s reality!

RECENT NEWS

  • Chronic Breathlessness Linked to Increased Financial Stress in Australian Survey
  • Deep Learning Electrocardiograms Underestimate Mortality Risk Under Clinical Case-Mix Shift
  • Female Traffic Deaths in Iran Linked to 121,929 Potential Foregone Births
  • Hybrid Deep Learning Model Detects Narcoleptic Sleep Irregularities With High Accuracy

Categories

  • Agriculture
  • Anthropology
  • Archaeology
  • Athmospheric
  • Biology
  • Biotechnology
  • Blog
  • Bussines
  • Cancer
  • Chemistry
  • Climate
  • Earth Science
  • Editorial Policy
  • Marine
  • Mathematics
  • Medicine
  • Pediatry
  • Policy
  • Psychology & Psychiatry
  • Science Education
  • Science News
  • Social Science
  • Space
  • Technology and Engineering

Subscribe to Blog via Email

Enter your email address to subscribe to this blog and receive notifications of new posts by email.

Join 5,151 other subscribers

© 2026 Scienmag - Science Magazine

Welcome Back!

Login to your account below

Forgotten Password?

Retrieve your password

Please enter your username or email address to reset your password.

Log In
No Result
View All Result
  • HOME
  • SCIENCE NEWS
  • CONTACT US

© 2026 Scienmag - Science Magazine

Discover more from Science

Subscribe now to keep reading and get access to the full archive.

Continue reading