Identification
Aperiam's computer vision algorithms know what chemical spatial patterns normally surround each amino acid in a folded protein.
Using our algorithms we analyze how well each amino acid fits in its local chemical environment and zero in on spots where the natural amino acid doesn't belong.
Surprisingly, we find that about three percent of the amino acids in a natural protein don't sit well in their folded context and that these amino acids are hidden from traditional sequence and energetics analyses.
swapping
We hone in on residues that don't fit in their current surroundings and ask our algorithms what residues would normally sit inside of that chemistry.
Typically, our algorithms suggest two or three different amino acids to try at each candidate site.
We construct those amino acid swaps and measure changes in each new protein's expression, stress tolerance, and catalytic turnover.
Normally, between 20% and 50% of our modifications improve these protein behaviors.
Combinations
Once we know which modifications strengthen our target protein in isolation we set about combining them together — tweaks that help in isolation are stronger together.
After testing roughly ten proteins that each contain several modifications we advance three to five of those proteins to pilot testing either internally or through a partner.
These derived proteins fold much better than their predecessors and are Aperiam's ultimate deliverable to our partner.
