Improved motif discovery. The rf-motifdiscovery module has been rewritten to mimic the DREME algorithm, further implementing the generation of motif logos without relying on WebLogo3. And much more! (7/n)

No more external dependency for structure plots. rf-fold now draws secondary structure plots natively, including an option to color bases on a continuous reactivity gradient. ViennaRNA RNAplot is no longer required. (6/n)

Four-base DMS analysis. The new rf-count and rf-norm now support four-base DMS experiments (developed by the Mustoe lab). (5/n)

Consensus folding across replicates. rf-fold now incorporates RNAalifold, restrained with the reactivity profiles of all replicates at once, yielding a single consensus structure rather than per-replicate models aggregated afterwards. (4/n)

Structure-informed alignment building and covariation analysis. A new module, rf-covalign, builds and refines structure-informed alignments of homologous sequences, and performs covariation analysis, wrapping Infernal and R-scape. (3/n)

Speed. Significant gains in runtimes across the board, between 3-25X, depending on the module. (2/n)