Fig. 6: Testing the predictive power of the CS + IF model.

a Comparison of experimentally measured and predicted apparent k_on, k_off, and the fold-change relative to unmethylated duplex (k_on fold-change = k_on(unmethylated)/ \({k}_{{{{{{\mathrm{on}}}}}},{{{{{\mathrm{m}}}}}}^{{{{{\mathrm{6}}}}}}{{{{{\mathrm{A}}}}}}}^{{{{{\mathrm{app}}}}}}\) and k_off fold-change = k_off(unmethylated)\ \({k}_{{{{{{\mathrm{off}}}}}},{{{{{\mathrm{m}}}}}}^{{{{{\mathrm{6}}}}}}{{{{{\mathrm{A}}}}}}}^{{{{{\mathrm{app}}}}}}\)) for RNA and DNA duplexes. Each point corresponds to a different duplex and/or experimental condition. All buffers contained 40 mM Na⁺, unless stated otherwise: (1) \({{{{{\mathrm{dsGGACU}}}}}}^{{{{{{\mathrm{m}}}}}}^6{{{{{\mathrm{A}}}}}}}\) at T = 65 °C, (2) at T = 55 °C, (3) with 3 mM Mg²⁺ at T = 65 °C; (4) \({{{{{\mathrm{dsHCV}}}}}}{{{{{\mathrm{m}}}}}}^{6}{{{{{\mathrm{A}}}}}}\) with 3 mM Mg²⁺ at T = 60 °C, (5) with 3 mM Mg²⁺ at T = 55 °C, (6) with 3 mM Mg²⁺ and 100 mM Na⁺ at T = 60 °C; (7) \({{{{{\mathrm{dsA6DNA}}}}}}^{{{{{{\mathrm{m}}}}}}^6{{{{{\mathrm{A}}}}}}}\) at T = 50 °C. Similar correlations were observed using RD simulation-based prediction method shown in Supplementary Fig. 9b. b Secondary structures of GS and ES in the apical loop of HIV-TAR with m⁶A35 (highlighted in red), showing the chemical structure of the m⁶A⁺-C bp. c Comparison of k_forward and k_backward for unmethylated TAR (A), experimentally measured (m⁶A exp.) and predicted (m⁶A calc.) for methylated TAR. d Secondary structures of GS and ES of methylated RREIIB. e Comparison of k_ex of unmethylated RRE (A), experimentally measured (m⁶A exp.), and predicted (m⁶A calc.) for methylated RRE. f Predicting the m⁶A-induced slowdown effect on \({k}_{{{{{{\mathrm{on}}}}}},{{{{{\mathrm{m}}}}}}^{{{{{\mathrm{6}}}}}}{{{{{\mathrm{A}}}}}}}^{{{{{\mathrm{app}}}}}}\) of 12-mers (see “Methods”) for m⁶A sites⁹ (orange) and random DNA (blue) in the mouse genome. Data in panels (a, c, e) were presented as mean values ±1 s.d. from Monte Carlo simulations (number of iterations = 500) for one CEST measurement as described in “Methods.” Source data for panels (a, c, e) are provided in the Source Data file.