Research · Biology
Two free, in-browser tools for biology papers: inspect AlphaFold protein folds in 3D, and turn a cell's gene counts into a Cell2Sentence with a transparent cell-type call. Nothing is uploaded.
Search a protein by name or UniProt accession and rotate the predicted fold. Confidence coloring (pLDDT) shows which parts of the ribbon you can trust before you drop a figure into LaTeX.
Paste one cell's gene counts. Ranked genes become a cell sentence; marker panels call the cell type; a log-log fit can reconstruct expression from rank. Encode, predict, and decode — all client-side.
AlphaFold predicted hundreds of millions of protein structures. This viewer loads a fold from AlphaFold DB so you can inspect secondary structure as a ribbon, spin the model, and judge confidence before you cite or figure it. It is the same 3D tool already on Bibby at /alphafold-viewer — kept as static HTML so it stays fast inside papers and the agent.
Cell2Sentence (C2S) ranks genes in a cell by expression and writes that ranking as a sentence of gene tokens. Full C2S language models need a GPU; this playground keeps the reversible rank transform and answers “what cell type is this?” by scoring the sentence against canonical marker panels in your browser. Use it to understand C2S sentences, not as a replacement for a trained C2S checkpoint on rare types.