Growing evidence shows that single‑nucleotide variants (SNVs), including cancer‑associated mutations, can reshape RNA secondary structures and thereby alter the binding of RNA‑targeting drug candidates. Because conventional profiling methods often fail to distinguish structural changes caused by genetic variation, researchers have lacked the tools needed to assess whether therapeutic compounds behave consistently across diverse patient genotypes. This limitation has become increasingly important as RNA‑targeting therapeutics advance in the clinic and as emerging evidence shows that SNVs can produce unexpected on‑ or off‑target effects.

BIVID‑MaP works by linking small‑molecule binding to a detectable signal during sequencing. The method uses a small molecule attached to a vinyl‑quinazolinone (VQ) group, which forms a covalent mark on nearby uridine bases when the molecule binds to its target RNA structure. During reverse transcription with the TGIRT enzyme, these marks cause the enzyme to skip exactly one nucleotide, creating a characteristic single‑nucleotide deletion. Deep sequencing can then pinpoint where these deletions occur. Because only one base is removed, the rest of the RNA sequence—including any genetic variant—is preserved, allowing each sequencing read to be assigned to the correct SNV. This approach avoids problems found in earlier methods, which often lose variant information due to stalled reverse transcription or confuse naturally occurring variants with chemically induced mutations.

G‑quadruplexes (G4s) —four‑stranded secondary structures formed from guanine‑rich sequences—are regulatory RNA structures implicated in controlling translation, splicing, and mRNA stability, and have been linked to several cancers. Using small molecules that bind RNA G4s, the researchers demonstrated that BIVID‑MaP sensitively detects binding signals and distinguishes interactions altered by SNVs. With G4‑binding compounds such as berberine and acridine derivatives, BIVID‑MaP identified variant‑dependent changes in binding. A striking example was a G→A mutation in the CD44 G4 element, which disrupted G4 formation and substantially reduced berberine binding. These differences were captured even when wild‑type and mutant RNAs were mixed, demonstrating the method's ability to resolve heterozygous variant‑specific interactions.

The research team next applied BIVID‑MaP to a large library of 1,621 5'-UTR RNAs from 283 cancer‑related genes, each containing a reference sequence and its corresponding tumor‑derived SNV. This screen revealed numerous variants that either enhanced or weakened small‑molecule binding. In DAXX, a 3A→G mutation increased G4 formation and strengthened berberine binding, whereas other substitutions produced the opposite effect. In ING2, a 128G→A mutation substantially reduced binding. These findings were validated using complementary assays, including affinity‑selection mass spectrometry and fluorescence‑based G4 detection, confirming that even subtle single‑nucleotide changes can alter RNA structural ensembles or ligand accessibility in ways that computational tools fail to predict.

BIVID‑MaP establishes a versatile platform for examining the structural consequences of genetic variation and their impact on RNA-ligand interactions. The method is scalable, compatible with large RNA libraries, and sensitive enough to detect interactions across a therapeutically relevant affinity range. By uncovering previously hidden variant‑dependent binding behaviors, BIVID‑MaP supports the development of RNA‑targeted therapeutics that remain effective across diverse human genotypes and helps minimize unexpected effects driven by naturally occurring mutations.

• doi: 10.1038/s41467-026-70097-9