What If You Could Screen Every Human Cell Type for Disease Signatures?
Public single-cell databases now contain over 140 million human cells spanning hundreds of tissues, diseases, and age groups. In principle, this is an unprecedented resource for drug discovery: you could ask questions like “which diseases show elevated stress response in astrocytes?” or “how does immune cell composition change with age across tissues?” In practice, nobody does this. The data is too heterogeneous. Metadata is inconsistent. Statistical modeling requires careful attention to batch effects, donor demographics, and tissue-specific confounds. Most researchers look at one dataset at a time. ...