Help
Help topics
You can provide sequences as plain text, FASTA files, or CSV files. For CSV uploads, the sequence column must be named 'sequence'.
RNA-FM embeddings are computed for uploaded sequences using a pretrained RNA-FM model. These embeddings are projected onto a shared 2D space using k-nearest-neighbor projection against reference siRNA, miRNA, and piRNA embeddings.
The Query Sequence panel allows you to paste a single RNA sequence and search for exact, fragment, and embedding-based nearest matches in the reference database. Matching sequences are highlighted on both the deep embedding plot and the interpretable feature plot, and a similarity-ranked results table is generated.
For each sequence, the tool computes length, GC percentage, GC skew, AT/AU skew, mononucleotide composition (MNC), and 2-mer and 3-mer frequencies.
Yes. In the Interpretable Features panel, you can select a subset of features. The reference dataset is re-projected using t-SNE based only on the selected features,
You can download exact, fragment, and embedding-based nearest matches sequence data with feature tables as a CSV file.