If you use Boltz1/2, BioEmu, Chai1, or other MSA-dependent models, you’re likely using our ColabFold server. Please be considerate! Avoid large submissions across many IPs instead generate the MSA locally. Our server is an old-timer from 2014 and can’t handle that load.

FreePMC articles… (mostly) no longer available outside the US.

A hill I will die upon: coriander has basically no taste, sub-herbaceous if I had to decide, so it tastes neither like citrus or soap; thus everyone is wrong.

Due process is the bedrock of any functioning judicial system. Just the act of carving exceptions ends any value or faith in the rule of law. The US democracy is teetering and we are ill equipped for the power vacuum that will follow.

Experimental validation of protein design models is critical, but what is perhaps more critical is doing so across research teams and lab environments. Same goes for beta testing the algorithms as well. The statistics of success, prior to release, should incorporate elements of usability.

'No one can expect individual scientists to make a stand, at professional and even personal danger (even though some are doing so),' writes Philip Ball. 'But prudence cannot mean timidity, nor can it risk looking like it.'

Substrate morphological preference in plastic depolymerases is poorly understood, but manipulating the surface charge of a highly active PETase, we were able to invert the preference, improve hydrolyitic activity, and discover a new mechanism for synergy along the way.

doi.org/10.1101/2025...

Perhaps at the moment, only a thought experiment, but would structure prediction training benefit from a more judicious selection of structured elements from PDBs; e.g. intentionally avoiding using structural elements supporting non-biological, crystal contacts.