Gene-centered identification of cis-regulatory islands highlights regulatory landscapes complementary to motif-centric approaches.

backgroundCis-regulatory elements constitute a fundamental layer of gene regulation, yet their computational identification has largely relied on transcription factor (TF)-centric frameworks that assume genome-wide background normalization and explicit TF binding models. While effective at the genome scale, such assumptions are less suitable for gene-centered analyses, where local sequence composition defines the relevant regulatory context.

resultsHere, we introduce a TF-independent observational framework for the gene-centered identification of cis-regulatory islands (GCIC), designed to characterize regulatory sequence organization based on local enrichment and diversity of short cis-regulatory sequence words derived from curated plant regulatory elements. Cis-regulatory islands identified by the GCIC framework (GCIC islands) are defined through the spatial overlap of independently enriched motif families, without relying on TF identity or genome-wide normalization. Application of the GCIC framework to the DROOPING LEAF (DL) locus in rice reveals discrete cis-regulatory islands, including one coinciding with a known intronic regulatory region, and highlights spatial patterns distinct from PWM-based motif scanning and clustering approaches. Genome-wide analysis shows that GCIC islands are broadly distributed but exhibit heterogeneous motif-family usage, with gene-level diversity and island-level reuse of motif-family combinations.

conclusionsWe demonstrate that cis-regulatory organization can be interpreted as a gene-centered property of sequence vocabulary usage within an observational framework. The GCIC framework provides a complementary representation of regulatory landscapes for gene-centered analyses, enabling locus-level interpretation of regulatory sequence organization beyond predictive inference.