The database of 200 million protein-structure predictions now includes homodimers, adding new biological relevance.
After years of creating highly specialized software, researchers used supercomputer clusters to finally solve the "100,000-body problem.