Processed DMS-MaPseq data for: Single-base substitutions switch RNA tertiary allostery between positive and negative coupling

Processed DMS-MaPseq data for the paper "Single-base substitutions switch RNA tertiary allostery between positive and negative coupling" (Oladeji, Kircher, Yesselman). Everything needed to regenerate every figure and number in the paper. The system. An engineered 129-nucleotide RNA "3D switch": an ATP aptamer joined to a GAAA tetraloop/tetraloop-receptor (TL/TLR) tertiary contact through two junctions, a kink-turn and a 4-1 junction. DMS-MaPseq reports both events at once. Four adenines report TL/TLR docking (three in the tetraloop, one in the receptor) and two adenines in the aptamer core report AMP binding. Higher mutation fraction means more exposed, so undocked or ligand-free. Single substitutions in the kink-turn flip the coupling between ligand binding and tertiary docking from positive to negative. Start here One file answers most questions: library/library_variants.csv — 7,147 variants x 38 columns, 3.6 MB. Opens in Excel, R or pandas. Two short documents make the rest legible: DATA_DICTIONARY.md defines every column, and CONSTRUCTS.md decodes the construct names. The same molecule appears as a lab code (C014K), a design name (atp_ttr_3) and a paper label ("Wild-type"), so CONSTRUCTS.md maps all three and records the traps. import pandas as pd df = pd.read_csv("library/library_variants.csv") # the switches df[df["class"].isin(["switch_on", "switch_off"])] # what drives switching: mutations in the kink-turn sheared pairs df.groupby("junction_class")["class"].value_counts() Each row carries the variant's sequence, structure, mutation string, DMS reactivity with and without AMP, and its allosteric classification. Useful columns: class — switch_on (AMP drives docking), switch_off (AMP undocks), always_on (docked, unresponsive), always_off (undocked, unresponsive), unclassified delta_tlr — the switching axis: TL/TLR reactivity without AMP minus with AMP. Positive means AMP increased docking gaaa_tlr_avg_wo_amp / _w_amp — the docking reporter in each condition atp_apt_avg_wo_amp / _w_amp — the AMP-binding reporter junction_class — kturn_only, 41_only, both, or neither (helix randomization only, the noise-floor set) kturn_pair_* — the geometry of each kink-turn signature pair after mutation, in standard Leontis-Westhof / Lilley numbering passed_read_filter — at least 2,000 aligned reads in both conditions. Only these rows are classified Contents library/ library_variants.csv (3.6 MB) — the table above library_combined.json (23.6 MB) — the same table plus per-nucleotide reactivity arrays for every variant in both conditions library_w_amp.json, library_wo_amp.json (20.8 MB each) — per-run library data for each condition library_metadata.json — schema version, pipeline settings, and the classification thresholds single-construct-runs.json (3.1 MB) — 923 samples: AMP, Mg2+ and temperature titrations of the wild type, the four representatives, the isolated aptamer, and two non-docking controls. Source for Figures 3 and 4 and every supplemental titration figure results/ — derived tables (CSV) coupling_summary.csv — four-state linked-equilibrium fits per construct: docking and binding constants, the coupling factor alpha, the coupling free energy, apparent Kd, with bootstrap confidence intervals coupling_mg_summary.csv, mg_coupling_summary.csv — Mg2+ docking midpoints with and without AMP, fold shifts, and the joint-fit coupling figure_4_model_selection.csv — two-state versus three-state Mg2+ docking selection: R2, delta-AICc, nested F-test figure_3_fit_parameters.csv, figure_4_fit_parameters.csv, figure_S_kd_comparison.csv — the fitted values plotted in those figures library_classification_*.csv — class counts, per-junction and per-mutation breakdowns, position enrichment, per-variant coupling factors references/ — six small JSON sidecars the analysis code reads at runtime, so constants are never hard-coded: canonical junction positions, the non-docking control reactivities that anchor the undocked threshold, the wild-type construct, and the representative constructs. config/ — experiment definitions: which sequencing runs and replicates make up each titration. The replicates field is the quality-control whitelist, the replicates actually used in the paper. DATA_DICTIONARY.md — every column of every file, the classification rules and thresholds, and the kink-turn numbering convention. CONSTRUCTS.md — the mapping between lab codes, design names and paper labels, plus four naming pitfalls. MANIFEST.txt and SHA256SUMS — file listing and checksums (shasum -a 256 -c SHA256SUMS). Using it with the analysis code The code is at github.com/YesselmanLabPublications/2026_atp_ttr_switch. Unpack this archive so that library/, results/ and references/ land in data/processed/ and config/ in data/config/, or let the repository's scripts/fetch_data.py do it. Then: switch-3d-paper references all # regenerate the reference sidecars python -m switch_3d_paper.cli.process_library # classify every variant switch-3d-paper analysis coupling # linked-equilibrium fits The repository README maps each manuscript figure to the notebook that builds it. How the classification works Variants are classified on two axes, the AMP-induced change in docking and the resting docking level, against thresholds derived from the data itself and recorded in library_metadata.json: sigma3 = 0.00287 (three times the standard deviation of the 3,590 helix-randomized variants, the experimental noise floor), theta_low = 0.00309 (1.1x the wild-type docking reactivity), theta_high = 0.01470 (midway between the helix-variant baseline and the non-docking controls). Of the 3,375 variants carrying a junction mutation that passed the read filter: 1,097 Always-On (32.5%), 1,900 unclassified (56.3%), 237 Switch-Off (7.0%), 120 Switch-On (3.6%), 21 Always-Off (0.6%). Related resources Raw sequencing readsNCBI BioProject PRJNA1483681899 SRA experiments. Demultiplexed reads for every condition in this deposit. Note: this BioProject is currently registered under the working title "The Design of Riboswitches without Secondary-Structure Rearrangements"; it is the raw-read archive for the study described here. Analysis codegithub.com/YesselmanLabPublications/2026_atp_ttr_switchInstalls, downloads this deposit, and regenerates every figure and number in the paper. This depositProcessed data: the classified variant library, the titration series, and all fitted parameters. Together these three cover the full chain from raw reads to published figure: reads at NCBI, processed data here, code on GitHub.