If you built the shape of a protein
the PDB does not have,
how far can you trust it?
In 2017 three researchers set out to find, by computer alone, a molecule that would bind the cancer target Ror1. They were stopped at the first step — no structure of this protein existed anywhere.
These 20 lessons follow how they built thousands of coordinates out of 388 letters of sequence, then measure the result with ProSA Z-scores, Ramachandran plots, Verify3D and estimated RMSD. Python runs right inside this page. There is nothing to install.
How you'll learn
A prediction server returns coordinates for anything you feed it. So six lessons on building are followed immediately by five on measuring.
Not a stand-in dataset. PROCHECK 67.4%, five disallowed residues, Drug Score 0.84 — the values as published.
Sequence identity, C-score, pairwise RMSD, error bars. This course teaches you to count what a paper leaves out as carefully as what it puts in.
Every lesson has a laboratory widget — twist a helix, move φ and ψ, slide the template coverage, re-rank the pockets yourself.