Artificial intelligence engine for generating candidate drugs

Abstract

An artificial intelligence engine for generating drug compounds is disclosed. In one embodiment, a method may include generating a biological context representation of a set of drug compounds. The biological context representation includes a first data structure having a first format. The method may also include translating, by the artificial intelligence engine, the first data structure having the first format to a second data structure having a second format. The method may also include generating, based on the second data structure having the second format, a set of candidate drug compounds. The method may also include classifying a candidate drug compound from the set of candidate drug compounds as a selected candidate drug compound.

What is claimed is: 
     
         1 . A method comprising:
 generating a biological context representation of a plurality of drug compounds, wherein the biological context representation comprises a first data structure having a first format;   translating, by an artificial intelligence engine, the first data structure having the first format to a second data structure having a second format;   generating, based on the second data structure having the second format, a plurality of candidate drug compounds; and   classifying a candidate drug compound from the plurality of candidate drug compounds as a selected candidate drug compound.   
     
     
         2 . The method of  claim 1 , wherein the biological context representation comprises, for each of the plurality of drug compounds, one or more relationships between or among:
 physical properties data,   chemical data,   biological data,   clinical outcome data, or   some combination thereof.   
     
     
         3 . The method of  claim 1 , wherein the translating the first data structure further comprises:
 converting the first data structure having the first format to the second data structure having the second format according to a specific set of rules executed by the artificial intelligence engine.   
     
     
         4 . The method of  claim 1 , wherein converting the biological context representation to the second data structure having the second format further comprises:
 obtaining a higher-dimensional vector from the biological context representation; and   compressing the higher-dimensional vector to a lower-dimensional vector, wherein the compressing is performed by a machine learning model trained to perform deep auto-encoding via a recurrent neural network configured to output the lower-dimensional vector.   
     
     
         5 . The method of  claim 4 , further comprising:
 training the machine learning model by using a second machine learning model to recreate the higher-dimensional vector,   wherein the second machine learning model is trained to perform a decoding operation to recreate the higher-dimensional vector, wherein the decoding operation is performed on the lower-dimensional vector.   
     
     
         6 . The method of  claim 1 , wherein:
 the translating is performed by a recurrent neural network,   the generating of the plurality of candidate drug compounds is performed by a generative adversarial network, and   the classifying of the candidate drug compound is performed by a classifier trained using supervised learning.   
     
     
         7 . The method of  claim 1 , further comprising:
 generating a plurality of views comprising a representation of a design space; and   causing the plurality of views to be presented on a computing device,   wherein the representation pertains to:
 antimicrobial activity, 
 immunomodulatory activity, 
 neuromodulatory activity, 
 cytotoxic activity, 
 or some combination thereof. 
   
     
     
         8 . The method of  claim 7 , wherein the design space is antimicrobial for prosthetic joint infections. 
     
     
         9 . The method of  claim 7 , wherein each view of the plurality of views presents an optimized sequence representing the selected candidate drug compound. 
     
     
         10 . The method of  claim 9 , wherein each view presents a topographical heatmap comprising the optimized sequence overlaid on indicators ranging from an at least one least active property to an at least one most active property. 
     
     
         11 . The method of  claim 1 , further comprising:
 performing one or more modifications pertaining to the biological context representation, the second data structure having the second format, or some combination thereof; and   using causal inference to determine whether the one or more modifications provide one or more desired performance results.   
     
     
         12 . The method of  claim 11 , wherein using causal inference further comprises:
 using counterfactuals to calculate alternative scenarios based on past actions, occurrences, results, regressions, regression analyses, correlations, or some combination thereof.   
     
     
         13 . The method of  claim 1 , further comprising:
 causing the selected candidate drug compound to be formulated.   
     
     
         14 . The method of  claim 1 , further comprising:
 causing the selected candidate drug compound to be created.   
     
     
         15 . The method of  claim 1 , further comprising:
 causing the selected candidate drug compound to be presented on a computing device.   
     
     
         16 . A tangible, non-transitory computer-readable medium storing instructions that, when executed, cause a processing device to:
 generate a biological context representation of a plurality of drug compounds, wherein the biological context representation comprises a first data structure having a first format;   translate the first data structure having the first format to a second data structure having a second format;   generate, based on the second data structure having the second format, a plurality of candidate drug compounds; and   classify a candidate drug compound from the plurality of candidate drug compounds as a selected candidate drug compound.   
     
     
         17 . The computer-readable medium of  claim 16 , wherein the biological context representation comprises, for each of the plurality of drug compounds, one or more relationships between or among:
 physical properties data,   chemical data,   biological data,   clinical outcome data, or   some combination thereof.   
     
     
         18 . The computer-readable medium of  claim 16 , wherein translating the biological context representation to the second data structure having the second format further comprises:
 obtaining a higher-dimensional vector from the biological context representation; and   compressing the higher-dimensional vector to a lower-dimensional vector, wherein the compressing is performed by a machine learning model trained to perform deep auto-encoding via a recurrent neural network configured to output the lower-dimensional vector.   
     
     
         19 . The computer-readable medium of  claim 18 , wherein the processing device further:
 trains the machine learning model by using a second machine learning model to recreate the higher-dimensional vector,   wherein the second machine learning model is trained to perform a decoding operation to recreate the higher-dimensional vector, wherein the decoding operation is performed on the lower-dimensional vector.   
     
     
         20 . The computer-readable medium of  claim 16 , wherein:
 the translating is performed by a recurrent neural network,   the generating of the plurality of candidate drug compounds is performed by a generative adversarial network, and   the classifying of the candidate drug compound is performed by a classifier trained using supervised learning.   
     
     
         21 . The computer-readable medium of  claim 16 , wherein the processing device further:
 generates a plurality of views comprising a representation of a design space,   wherein the representation pertains to:
 antimicrobial activity, 
 immunomodulatory activity, 
 neuromodulatory activity, 
 cytotoxic activity, 
 or some combination thereof. 
   
     
     
         22 . The computer-readable medium of  claim 21 , wherein the processing device further:
 causes each view of the plurality of views to be presented, wherein each such view comprises an optimized sequence representing the selected candidate drug compound, and   causes each view of the plurality of views to be presented, wherein each such view comprises the optimized sequence overlaid on indicators ranging from an at least one least active property to an at least one most active property.   
     
     
         23 . The computer-readable medium of  claim 16 , wherein the processing device further:
 performs one or more modifications pertaining to the biological context representation, the second data structure having the second format, or some combination thereof and   uses causal inference to determine whether the one or more modifications provide a desired performance result.   
     
     
         24 . The computer-readable medium of  claim 23 , wherein the processing device employs causal inference to further:
 use counterfactuals to calculate alternative scenarios based on past actions, occurrences, results, regressions, regression analyses, correlations, or some combination thereof.   
     
     
         25 . A system comprising:
 a memory device storing instructions; and   a processing device communicatively coupled to the memory device, the processing device executes the instructions to:
 generate a biological context representation of a plurality of drug compounds, wherein the biological context representation comprises a first data structure having a first format; 
 translate the first data structure having the first format to a second data structure having a second format; 
 generate, based on the second data structure having the second format, a plurality of candidate drug compounds; and 
 classify a candidate drug compound from the plurality of candidate drug compounds as a selected candidate drug compound. 
   
     
     
         26 . A computing device comprising:
 a display screen;   a memory device storing instructions; and   a processing device communicatively coupled to the memory device and the display screen, the processing device executes the instructions to:
 receive, from an artificial intelligence engine, a candidate drug compound generated by the artificial intelligence engine; 
 generate a view comprising the candidate drug compound overlaid on a representation of a design space, wherein the view presents a topographical heatmap of the representation of the design space, and the topographical heatmap includes the candidate drug compound overlaid on indicators ranging from an at least one least active property to an at least one most active property; and 
 present the view on the display screen. 
   
     
     
         27 . A method comprising:
 generating a biological context representation of a plurality of drug compounds, wherein the biological context representation comprises a first data structure having a first format;   translating, by an artificial intelligence engine, the first data structure having the first format to a second data structure having a second format;   generating, based on the second data structure having the second format, a plurality of candidate anti-biofilm compounds; and   classifying a candidate anti-biofilm compound from the plurality of candidate anti-biofilm compounds as a selected candidate anti-biofilm compound.   
     
     
         28 . A method comprising:
 generating a biological context representation of a plurality of drug compounds, wherein the biological context representation comprises a first data structure having a first format;   translating, by an artificial intelligence engine, the first data structure having the first format to a second data structure having a second format;   generating, based on the second data structure having the second format, a plurality of candidate anti-cancer compounds; and   classifying a candidate anti-cancer compound from the plurality of candidate anti-cancer compounds as a selected candidate anti-cancer compound.   
     
     
         29 . A method comprising:
 generating a biological context representation of a plurality of drug compounds, wherein the biological context representation comprises a first data structure having a first format;   translating, by an artificial intelligence engine, the first data structure having the first format to a second data structure having a second format;   generating, based on the second data structure having the second format, a plurality of candidate antimicrobial compounds; and   classifying a candidate antimicrobial compound from the plurality of candidate antimicrobial compounds as a selected candidate antimicrobial compound.