You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Pinned tt-bio moves to 0.7.0. Every fold time on every card, the site and
this handout is re-measured on that release; the numbers barely move, and
the 0.6.7 Protenix-v2 accuracy fixes are not visible on these targets
because all seven fold without an MSA.
The weight-artifact API the weights postinst and the runner both call
moved into tt_bio.weights upstream; both are ported, and the packaging
contract test now checks the artifact key as well as the function.
A finished structure now holds the screen for as long as the NEXT fold can
afford rather than a flat two seconds, capped per target by its measured
time to first coordinates: about seven seconds ahead of the long targets,
unchanged ahead of the short ones, whose whole collapse a flat hold would
have suppressed.
The cartoon draws nucleic acids instead of dropping them. A protein/DNA
complex used to render the protein and silently omit the DNA, and a pure
DNA or tRNA fold fell back to the old tube renderer by way of an
exception. Same picture as before for those two, now drawn on purpose.
A booth with more than one chip comes up showing all of them. --quad still
forces the grid on a single-chip booth and the new --solo forces one large
protein; Q toggles either way at runtime, as before.