Protein structures from amino-acid sequences using neural networks
The present disclosure provides for systems and methods for generating and displaying a three dimensional map of a protein sequence. An exemplary method can provide for using deep learning models to predict protein folding and model protein folding using three dimensional representations. The method more effectively exploits the potential of deep learning approaches. The method approach overall involves three stages—computation, geometry, and assessment.
1. A system comprising:
a display;
a memory containing non-transitory machine-readable medium comprising machine executable code having stored thereon instructions;
a control system coupled to the memory comprising one or more processors, the control system configured to execute the machine executable code to cause the one or more processors to:
receive molecular data comprising a set of amino acid residues of a protein;
process the set of amino acid residues using a recurrent geometric network;
output a three dimensional map of the protein comprising three dimensional Cartesian coordinates; and
display the three dimensional map on the display; and
wherein the control system is further configured to compute deviations between predicted and experimental structures using a distance-based root mean square deviation metric.
2. The system of claim 1 , wherein processing the set of amino acid residues using a recurrent geometric network further comprises:
determining an internal state for each residue of the set of amino acid residues;
integrating the internal state with the states of adjacent amino acid residues for each residue of the set of amino acid residues;
determining, based on the integrated internal states, geometric units from predicted torsional angles; and
translating geometric units to Cartesian coordinates.
3. The system of claim 1 , wherein the control system is further configured to store, in a memory, the three dimensional map of the protein comprising three dimensional Cartesian coordinates.
4. The system of claim 1 , wherein the control system uses a recurrent geometric network.
5. The system of claim 1 , wherein control system network is optimized to minimize the distance-based root mean square deviation metric.
6. The system of claim 1 , wherein the control system generates a fully differentiable map extending from sequence to structure.
7. A method of using a control system comprising:
receiving molecular data comprising a set of amino acid residues of a protein;
processing the set of amino acid residues using a recurrent geometric network;
outputting a three dimensional map of the protein comprising three dimensional Cartesian coordinates; and
displaying the three dimensional map on the display; and
wherein the control system is further configured to compute deviations between predicted and experimental structures using a distance-based root mean square deviation metric.
8. The method of claim 7 , wherein processing the set of amino acid residues using a recurrent geometric network further comprises:
determining an internal state for each residue of the set of amino acid residues;
integrating the internal state with the states of adjacent amino acid residues for each residue of the set of amino acid residues;
determining, based on the integrated internal states, geometric units from predicted torsional angles; and
translating geometric units to Cartesian coordinates.
9. The method of claim 7 , wherein the control system is further configured to store, in a memory, the three dimensional map of the protein comprising three dimensional Cartesian coordinates.
10. The method of claim 7 , wherein the control system uses a recurrent geometric network.
11. The method of claim 7 , wherein the control system is optimized to minimize the distance-based root mean square deviation metric.
12. The method of claim 7 , wherein the control system generates a fully differentiable map extending from sequence to structure.