Machine learning enabled prediction of molecular structures and properties

A computer-implemented method, comprising:

• receiving a molecular structure file specifying an initial three-dimensional structure of a protein molecule comprising a sequence of amino acid residues;
• determining, based at least on the molecular structure file, a representation of the protein molecule that includes a plurality of frames for each amino acid residue in the sequence of amino acid residues,
• wherein the plurality of frames for each amino acid residue includes a first set of frames specifying a geometric state a backbone of the amino acid residue, and
• wherein the first set of frames includes a rotation matrix specifying the rotation of the backbone of the amino acid residue or a displacement vector specifying the translation of the backbone of the amino acid residue, and
• wherein the representation of the protein molecule further includes, for each position in a sequence of amino acid residue forming the protein molecule, a probability distribution across a set of possible amino acid residues occupying the position;
• and
• generating a three-dimensional structure of the protein molecule by at least applying a design computation model to modify the representation of the protein molecule, wherein the design computation model modifies the plurality of frames while also modifying the probability distribution such that a probability of the position being occupied by each possible type of amino acid residue sums to 1.

A non-transitory computer readable medium storing instructions, which when executed by at least one data processor coupled with a computing system, result in operations comprising:

• receiving a molecular structure file specifying an initial three-dimensional structure of a protein molecule comprising a sequence of amino acid residues;
• determining, based at least on the molecular structure file, a representation of the protein molecule that includes a plurality of frames for each amino acid residue in the sequence of amino acid residues,
• wherein the plurality of frames for each amino acid residue includes a first set of frames specifying a geometric state a backbone of the amino acid residue, and
• wherein the first set of frames includes a rotation matrix specifying the rotation of the backbone of the amino acid residue or a displacement vector specifying the translation of the backbone of the amino acid residue, and
• wherein the representation of the protein molecule further includes, for each position in a sequence of amino acid residue forming the protein molecule, a probability distribution across a set of possible amino acid residues occupying the position;
• and
• generating a three-dimensional structure of the protein molecule by at least applying a design computation model to modify the representation of the protein molecule, wherein the design computation model modifies the plurality of frames while also modifying the probability distribution such that a probability of the position being occupied by each possible type of amino acid residue sums to 1.