Discover clusters of short RNA motifs (tetramers) — multivalent RNA motifs (MRMs) in the paper — that are positionally enriched around alternatively spliced exons regulated by an RNA-binding protein (RBP), and read the resulting RNA splicing maps.

Method reference: Cereda M. et al. RNAmotifs: prediction of multivalent RNA motifs that control alternative splicing. Genome Biology 2014;15(1):R20.

Grieco M, Becchi T, Boscagli G, Priante F, Caizzi L, Pozzoli U, Cereda M. RNAMaRs: an interpretable framework for inferring multivalent RNA Motifs and cognate Regulators of Splicing. doi:10.64898/2026.01.31.703040.

Tutorial pages

• Input Format — file layout and the critical dIRank convention
• Parameters — every flag explained, with tuning guidance
• Output & Interpretation — what each file is, reading the splicing map
• Examples — copy-paste command recipes
• Troubleshooting — FAQ and common pitfalls
• RNAmotifs-MaRs — coupling with eCLIP for MRM–RBP association scores

🚧 Intron retention (--event-type RI), secondary structure (--structure) and conservation (--conservation) are work-in-progress features — see the notes on the relevant pages.

---

What RNAmotifs does

When you knock down an RBP and run RNA-seq, some cassette exons are included more (enhanced) and some less (silenced). RNAmotifs asks:

Which short sequence motifs, clustered at which positions around those exons, distinguish the regulated exons from unaffected (control) exons?

It takes three exon sets — enhanced, silenced, control — scans the sequence around each exon's splice sites for every k-mer (tetramers by default), finds positions where a motif forms a multivalent cluster (an MRM; several copies close together), and tests whether each motif's clusters are enriched in the regulated sets vs control. The output is an RNA splicing map plus a statistically filtered list of enriched motifs.

---

2-minute quick start

# Build binaries (see README "Installation"):  cmake -B build && cmake --build build
# Download your genome:                         bash genomes/hg19.download.sh

./rnamotifs my_exons.txt \
    --name MYRBP \
    --genome hg19 \
    --bootstraps 1000 \
    --cores 8

Writes a results folder under results/MYRBP/<timestamped-run>/ with the RNA splicing-map PDF, the enriched-tetramer table, and per-region/per-group CSVs (see Output & Interpretation).

Tip: use --bootstraps 1000 for everyday runs (fast). --bootstraps 10000 is ~10× slower for a lower empirical-p floor that rarely changes the called set.

---

How the algorithm works (in five steps)

Understanding these makes every parameter intuitive.

1. Define regions around each exon's two splice sites — into the exon (--in-exon bp) and into the flanking introns (--in-intron bp). For a skipped exon: R1 = upstream intron, R2 = exon body, R3 = downstream intron.
2. Find motif clusters. For each k-mer, slide along each region and count copies within a half-window (--half-window). Where the local count (cluster height) reaches --min-height, the position is "occupied" — this is what multivalent means.
3. Positional occupancy per group — fraction of enhanced / silenced / control exons with a cluster at each position (--pth binarises occupancy). Plotting this is the splicing map.
4. Enrichment test — Fisher's exact test, regulated vs control, per region; must beat --p-fisher.
5. Empirical FDR — resample control exons --bootstraps times for an empirical p-value; must also beat --p-empirical.

So hw/min-height/pth define what counts as a cluster; in-exon/in-intron define where you look; p-fisher/p-empirical/bootstraps define how strict the filter is.

---

Supported genomes

The matching genome must be downloaded first (genomes/<assembly>.download.sh).