tooluniverse-noncoding-rna

Query ENCORI/starBase with ENCORI_get_miRNA_targets using the miRNA name to get CLIP-seq-supported targets ranked by experimental evidence. For literature-validated targets, search PubMed for reporter assay or CLIP-seq studies, since computational predictions alone have high false positive rates.

Search LNCipedia for transcript details, sequence, and conservation, then query PubMed with the lncRNA name plus "mechanism" or "function". lncRNA mechanisms are determined experimentally, so literature evidence is essential rather than name-based inference.

It uses miRBase for miRNAs, LNCipedia for lncRNAs, RNAcentral aggregating 40+ databases for all ncRNA types, Rfam for RNA families, ENCORI for miRNA-target interactions, and DisGeNET plus GTEx for disease and expression context.

TargetScan has no REST API, so predictions require downloading and processing its bulk data files locally with pandas. miRTarBase blocks programmatic access, so use its bulk export or R data package, or fall back to ENCORI and PubMed searches.

Even the best prediction algorithms have over 50% false positive rates, so only experimentally validated targets from reporter assays or CLIP-seq should be reported as findings. Predictions from TargetScan or miRanda are hypothesis-generating only.

Use the RNA-seq DESeq2 skill for mRNA differential expression analysis and the CRISPR screen analysis skill for screen data. This skill covers only non-coding RNA identification, targets, and functional interpretation.