An atlas of centromere architecture reveals recurrent evolutionary dynamics
Year: 2026
Type: dataset
Abstract: SUPPORTING DATA AND ANALYSIS TABLES=================================== ##### Supplementary Data 1 [linked to 'data_bundles_annotations.tar.gz']. For each of the 325 genomes we provide a zip-bundle of files containing:a) CAP output plot(s).b) Filtered EDTA transposon annotation in gff format.c) Filtered Helixer gene annotation in gff format.d) Filtered TRASH tandem repeat annotation in gff format.e) The associated genome fai index file.f) A README file explaining the files. ##### Supplementary Data 2 [linked to 'satellite_data_sheets.zip']. Candidate centromeric satellite families. For each satellite repeat family that CAP classified as a centromere candidate, a single file is provided with three plots. On the left, each repeat is plotted as a single horizontal black line, with ordering by the length of each repeat, then by chromosome and coordinate when repeats have the same length. The location of CG, CC, GG, and GC dinucleotides are shown by the indicated colours. The middle plot is the same as on the left, but the repeats are ordered by chromosome, then by coordinate. A key is presented to indicate chromosome number to the left, starting at 1 from the bottom of the plot. The plot on the right analyses the repeat family's conformity to Zipf's law (the r2 value is printed in the plot title). ##### Supplementary Data 3 [linked to 'transposon_data_sheets.zip']. Centrophilic transposon atlas. Compilation of transposon data shown with mulitple plots for every species with CAP-predicted satellite arrays. Plot description:a. Table with general information for every genome, including the species Tree of Life Identifier (ToLID), genome size, centromeric satellite family, and number and total length of non-satellite interruptions surveyed for transposons (minimum interruption length 250 bp).b. Composition (in bp) of every centromere by centromeric satellites (red), transposons (blue), unannotated DNA (grey), and other low-copy satellites that have not been classified as part of the centromeric arrays (green). Centromeres are sorted by length, with the box plot at the bottom showing the distribution of centromere lengths.c. As for b, but with box plots showing the contribution (in bp) of each feature (centromeric satellites, transposons, unannotated DNA, other non-centromeric satellites) for all centromeres combined.d. Proportion of transposon annotation that was classified as intact (dark blue) versus fragmented (light blue) inside and outside the centromeres.e. LTR identity of intact LTR elements inside and outside the centromeres.f. Graphs showing the length of centromere space that was allocated to centromeric satellites and interruptions; with interruptions further resolved into transposons, unannotated DNA, and other non centromeric satellites; and transposons further resolved into Class_I_LTR (LTR retrotransposons and retroviruses), Class_I_TPRT (non-LTR retrotransposons such as LINEs that undergo target-primed reverse transcription), Class_I_other (Penelope and Tyrosine Recombinase elements), and Class_II_DNA (elements that have terminal inverted repeats such as Mutator and CACTA, and also Helitrons). 'TE_unclass' and 'repeat_region' refer to annotations of the EDTA pipeline that were not reassigned to a specific transposon class by our dedicated reannotation pipeline (see Methods).g. As for f, but showing the composition for each centromere separately. The proportion refers to the length of the centromere in relation to the whole chromosome, while the karyotype shows the position and length of every centromere. All karyotypes are in scale. The first row shows the composition of all centromeres combined. ##### CENH3_analysis [linked to 'CENH3 analysis.zip'] CENH3_analysis/---------------Contains the CENP-A/CENH3 protein phylogeny and the sequence analyses built on it (split entropy, GroupSim specificity, terminal branch lengths). CENH3_CENP-A_422.fastaThe 422 curated CENP-A/CENH3 protein sequences (unaligned, full length) used throughout the study. H3_canonical_897.fastaThe 897 canonical histone H3 protein sequences (unaligned, full length), the comparison group for CENP-A. cenpa430_H3_archaea10.aligned.clipkit.325sp.fastaThe master protein alignment: 1,329 sequences = 422 CENP-A + 897 H3 + 10 archaeal-histone outgroup sequences (MAFFT, trimmed with ClipKit). This is the input to the CENH3 tree and to all column-wise analyses below. CENH3_H3_archaea.VTG4.treefileMaximum-likelihood CENP-A/CENH3 + H3 + archaea protein tree (1,329 tips), IQ-TREE2, model VT+G4 (ModelFinder/BIC), -bnni, 1000 ultrafast bootstraps. CENH3_H3_archaea.VTG4.contreeThe corresponding bootstrap consensus tree. CENH3_H3_archaea.VTG4.iqtreeIQ-TREE run report: selected model, likelihoods, and branch support. CENH3_H3_archaea.VTG4.logIQ-TREE run log. CENH3_analysis/entropy_and_groupsim/------------------------------------Column-wise analyses on the gap-0.85-trimmed alignment. groups_cenpa_vs_h3.txtFull membership of the CENP-A vs H3 split (two lines, comma-separated IDs); the 10 archaeal outgroup sequences are excluded from both groups. groups_satellite_vs_transposon.txtArchitecture split of CENP-A by the host species' centromere type: Satellite = 196, Transposon = 130 (holocentric/cryptic/unknown excluded). split_entropy_cenpa_vs_h3_gap085.tsvPer-position Shannon entropy per group (pos, all, CENPA, H3), 147 positions. groupsim_cenpa_vs_h3_raw_gap085.txtRaw GroupSim output for CENP-A vs H3 (per column: index, score, per-group residue composition; trimmed alignment appended). Unweighted. groupsim_cenpa_vs_h3_cladeweighted_gap085.tsvClade-weighted CENP-A-vs-H3 GroupSim (pos, groupsim, z, sig), 147 positions; 6 significant positions (max z 3.56). groupsim_satellite_vs_transposon_raw_gap085.txtRaw GroupSim output for Satellite vs Transposon. groupsim_satellite_vs_transposon_cladeweighted_gap085.tsvClade-weighted Satellite-vs-Transposon GroupSim, 218 positions; 0 significant (max z 1.88). CENH3 analysis/terminal_branch/-------------------------------Terminal-branch-length (TBL) comparison. terminal_branch_per_tip_cenpa_vs_h3.tsvPer-tip TBL distribution (one row per sequence; 1,319 tips = 422 CENP-A + 897 H3). terminal_branch_per_species_cenpa_vs_h3.tsvPer-species median TBL for each group, for the 262 species with >=1 CENP-A and >=1 H3. terminal_branch_stats.tsvSummary and test: paired Wilcoxon signed-rank on the 262 species. ##### Phylogeny [linked to 'Phylogeny.zip'] Phylogeny/----------The time-calibrated 325-species species tree, its inference and calibration inputs, ancestral-state reconstruction of centromere architecture, satellite-similarity decay, and per-trait phylogenetic signal. species_tree_325sp_chronos_lambda01_62cal.nwkSpecies tree: 325 tips, branch lengths in millions of years (Mya), dated with chronos (correlated-rates, dating lambda = 0.1) from 62 TimeTree calibrations; root age ~1,475 Mya. Used by all downstream analyses. FastSpeciesTree_IQTREE.treefileThe underlying maximum-likelihood species tree (uncalibrated; substitutions/site, with ultrafast-bootstrap support) from FastSpeciesTree + IQ-TREE. FastSpeciesTree_IQTREE.contreeBootstrap consensus of the ML species tree. FastSpeciesTree_IQTREE.iqtree / .logIQ-TREE run report (model, partitions, likelihoods, support) and run log. FastSpeciesTree_IQTREE.best_model.nex / .best_scheme.nexModelFinder best models and the partitioning scheme. FastSpeciesTree_trimmed_supermatrix.fastaThe trimmed concatenated BUSCO pseudo-alignment (supermatrix) = IQ-TREE input. FastSpeciesTree_partitions.txtPartition definitions for the trimmed supermatrix. FastSpeciesTree_pipeline.logFull FastSpeciesTree log from the start: DIAMOND database build, per-proteome BLAST, and BUSCO gene selection (194 genes retained across 325 species), through to the IQ-TREE call. calibration_points_62.tsvThe 62 node calibrations used for dating (clade, node, taxon pair, min/max Mya, calibrated age, TimeTree median/adjusted/range, source). All TimeTree-derived. Phylogeny/ASR/--------------Ancestral-state reconstruction of centromere architecture (3 states: H = holocentric, Sat = satellite, Trans = transposon) with phytools::fitMk (ML; models compared by AICc and Akaike weights). model_comparision/model_comparison_3state_chronos_lambda01.tsv7-model AICc comparison (ER, SYM, ARD, ARD_irrevH, 3 reduced directional) on the chronos lambda=0.1 tree, per dataset (full, Metazoa, Viridiplantae). Best: ARD_irrevH (full/Metazoa), SYM (Viridiplantae); Sat<->Trans bidirectional. model_comparision/transition_rates_3state.tsvFitted instantaneous transition rates (per My) for ER/SYM/ARD/ARD_irrevH, per dataset and directed transition — the numbers behind the transition-rate flow diagram. annotation/centromere_architecture_annotation_325sp.tsvPer-species centromere-architecture annotation (tip symbol/colour/class). annotation/asr_input_states_277tips.tsvThe exact ASR input: each tip's architecture and 3-state label; 277 used + 48 pruned (11 mixed + 37 cryptic) = 325. Phylogeny/half_life/--------------------Decay of centromeric-satellite sequence similarity with divergence time (BLASTN, all-vs-all), fitted as sim(t) = A*exp(-lambda*t) + C; half-life = ln2/lambda. seqsim_blastn_allpairs_chronos_lambda01_nov2026.tsvAll 13,203 species-pair BLASTN comparisons (species pair, divergence time, mean % identity, fraction of satellites with a hit, hit count, clade group). seqsim_blastn_nodeaveraged_chronos_lambda01_nov2026.tsvThe all-pairs data aggregated to one point per MRCA node (157 node points). halflife_perclade_nov2026.tsvFitted per-clade result. Phylogeny/pagel/---------------- pagels_lambda_per_trait.tsvPagel's lambda per centromere/genome trait on the species tree (geiger::fitContinuous, ML; cross-validated with nlme::gls/corPagel; lambda bounded [0,1]; profile-likelihood 95% CIs). Columns: trait, n, lambda, 95% CI, gls cross-validation lambda, logL, logL(lambda=0), LR statistic ce
Language: English
Subfield: Genetics
Domain: Life Sciences
Open Access status: green
OpenAlex ID: W7225754099
