Method and apparatus for identification of biomolecules

Abstract

The present disclosure presents methods, systems, and devices for dentifying new molecules directly from biological sequence information, with at least one of a desired bioactivity profile, functional attribute, biochemical reactivity, biological impact, pharmacological characteristic or therapeutic effect. The present disclosure further includes analyzing, at the processor, data features of biological sequence information and other data sources, including a feature-definition set by processing, using one or more bioinformatic techniques, computational algorithms, or methods of statistical machine learning, data sources relating to biological or chemical molecules, including biomolecules, including but not limited to peptides, having desired physical or chemical characteristics, bioactivities, functional attributes, biological impacts, pharmacologic properties or therapeutic effects.

1 .- 33 . (canceled) 
     
     
         34 . A method of identifying, on an electronic device comprising a processor and a memory, directly from biological sequence information, biomolecules having a desired bioactivity, the method comprising:
 (a) processing, at the processor, data sources relating to a desired bioactivity, including biological sequence information, using a method of statistical machine learning,   (b) analyzing, at the processor, using the method of statistical machine learning, a first set of one or more physical, chemical, biological, pharmacological or functional attributes (each, a “distinguishing feature”), wherein each distinguishing feature distinguishes validated biomolecules having a desired bioactivity profile comprising bioactivities, functional attributes, biological impacts, pharmacologic properties or therapeutic effects, from rejected biomolecules lacking the desired bioactivity profile, wherein the distinguishing feature is a nucleotide sequence motif, a nucleotide or nucleotide sequence encoding a peptide sequence or structure motif with the desired bioactivity profile, or a nucleotide sequence encoding a RNA sequence or structure motif with the desired bioactivity profile,   (c) deriving at the processor, an improved set of distinguishing features that, as compared to the first set of distinguishing features, more accurately distinguishes validated biomolecules with the desired bioactivity profile from rejected biomolecules lacking the desired bioactivity profile, by:
 (i) using the algorithm from step (b) to search, at the processor, genome sequence data from databases for candidate biomolecules predicted to exhibit the desired bioactivity profile; 
 (ii) compiling from the genome sequence data, at the processor, a list of candidate biomolecules based on step (c)(i) 
 (iii) producing at a biomolecule synthesis system a random subset of the candidate biomolecules; 
 (iv) measuring, at a bioassay evaluation device, the presence and level of the desired bioactivity profile, and scoring the candidates based upon the relative measures or performance of the tested subset of candidate biomolecules at the bioassay evaluation device; 
 (v) determining at the processor, using the method of statistical machine learning, the features that distinguish candidate biomolecules having the bioactivity profile from tested candidate biomolecules lacking the bioactivity profile, based upon the results of (c)(iv), and using this information to modify the algorithm in step (c)(i); and 
   (d) repeating steps (i) to (v) at least once, at the processor, to further refine the features that distinguish candidate biomolecules having the desired bioactivity profile from those candidate biomolecules that lack, or have a reduced or suboptimal bioactivity profile,   
       wherein upon the final repetition of step (c), the candidate biomolecules compiled in step (c)(ii) identify biomolecules having, or most likely to exhibit, the desired bioactivity profile and identify biomolecules for use as or further development as a drug; pesticide; herbicide; antimicrobial; or a molecule for influencing crop plant or animal performance, biofuel production, or detoxification of polluted environments. 
     
     
         35 . The method of  claim 34 , wherein the method of statistical machine learning is selected from the group consisting of latent class analysis, Bayesian inference, support vector machines, directed acyclic graphical models, and artificial neural networks, and a combination thereof. 
     
     
         36 . The method of  claim 34 , wherein the bioactivity profile comprises at least one of drug efficacy data and drug safety data. 
     
     
         37 . The method of  claim 36 , wherein the bioactivity profile comprises at least one of gene expression data, mammalian LD50 data, metabolic data, pharmacokinetic data, excretion or clearance data, liver toxicity data, absorption data, membrane permeability data, cellular localization data, and small molecule analog data. 
     
     
         38 . The method of  claim 36 , wherein the bioactivity profile comprises drug efficacy data and drug safety data. 
     
     
         39 . The method of  claim 38 , wherein the bioactivity profile further comprises at least one of gene expression data, mammalian LD50 data, metabolic data, physiological transport dynamics, pharmacokinetic data, water solvation data, absorption data, excretion or clearance data, liver toxicity data, membrane permeability data, cellular localization data, small molecule analog data, diffusion coefficient, and oral bioavailability data. 
     
     
         40 . The method of  claim 34 , wherein the biological sequences are selected from the group consisting of DNA nucleotide sequences, RNA nucleotide sequences, amino acid sequences, chemically modified DNA nucleotide sequences, biologically modified DNA nucleotide sequences, chemically modified RNA nucleotide sequences, biologically modified RNA nucleotide sequences, chemically modified amino acid sequences, biologically modified amino acid sequences, chemically modified protein sequences, and biologically modified protein sequences. 
     
     
         41 . The method of  claim 34 , wherein the desired bioactivity is an effect selected from the group consisting of biochemical, biophysical, pharmacological, therapeutic, antimicrobial, cytotoxic, antitumor, antiproliferative, and antineoplastic. 
     
     
         42 . The method of  claim 34 , wherein the distinguishing features of the validated biomolecules with the desired bioactivity profile are associated with a chemical physical, biological, pharmacological, or clinical functionality of an amino acid sequence, a RNA sequence, or a DNA sequence, constituting or coding for a molecule with anti-bacterial, anti-viral, anti-fungal, anti-parasitic, or anti-pathogen activity. 
     
     
         43 . A method of identifying, on an electronic device comprising a processor and a memory, directly from biological sequence information, biomolecules having a desired bioactivity, the method comprising:
 (a) processing, at the processor, data sources relating to a desired bioactivity, including biological sequence information, using a method of statistical machine learning,   (b) analyzing, at the processor, using the method of statistical machine learning, a first set of one or more physical, chemical, biological, pharmacological or functional attributes (each, a “distinguishing feature”), wherein each distinguishing feature distinguishes validated biomolecules having a desired bioactivity profile comprising bioactivities, functional attributes, biological impacts, pharmacologic properties or therapeutic effects, from rejected biomolecules lacking the desired bioactivity profile, wherein the distinguishing feature is a nucleotide sequence motif, a nucleotide or nucleotide sequence encoding a peptide sequence or structure motif with the desired bioactivity profile, or a nucleotide sequence encoding a RNA sequence or structure motif with the desired bioactivity profile,   (c) deriving at the processor, an improved set of distinguishing features that, as compared to the first set of distinguishing features, more accurately distinguishes validated biomolecules with the desired bioactivity profile from rejected biomolecules lacking the desired bioactivity profile, by:
 (i) using the algorithm from step (b) to search, at the processor, genome sequence data from databases for candidate biomolecules predicted to exhibit the desired bioactivity profile; 
 (ii) compiling from the genome sequence data, at the processor, a list of candidate biomolecules based on step (c)(i) 
 (iii) producing at a biomolecule synthesis system a random subset of the candidate biomolecules; 
 (iv) measuring, at a bioassay evaluation device, the presence and level of the desired bioactivity profile, and scoring the candidates based upon the relative measures or performance of the tested subset of candidate biomolecules at the bioassay evaluation device; 
 (v) determining at the processor, using the method of statistical machine learning, the features that distinguish candidate biomolecules having the bioactivity profile from tested candidate biomolecules lacking the bioactivity profile, based upon the results of (c)(iv), and using this information to modify the algorithm in step (c)(i); and 
   (d) repeating steps (i) to (v) at least once, at the processor, to further refine the features that distinguish candidate biomolecules having the desired bioactivity profile from those candidate biomolecules that lack, or have a reduced or suboptimal bioactivity profile,   
       wherein upon the final repetition of step (c), scoring the candidate biomolecules in step (c)(iv) identifies biomolecules having the desired bioactivity profile and identifies biomolecules for use as or further development as a drug; pesticide; herbicide; antimicrobial; or a molecule for influencing crop plant or animal performance, biofuel production, or detoxification of polluted environments. 
     
     
         44 . The method of  claim 43 , wherein the method of statistical machine learning is selected from the group consisting of latent class analysis, Bayesian inference, support vector machines, directed acyclic graphical models, and artificial neural networks, and a combination thereof. 
     
     
         45 . The method of  claim 43 , wherein the bioactivity profile comprises at least one of drug efficacy data and drug safety data. 
     
     
         46 . The method of  claim 45 , wherein the bioactivity profile comprises at least one of gene expression data, mammalian LD50 data, metabolic data, pharmacokinetic data, excretion or clearance data, liver toxicity data, absorption data, membrane permeability data, cellular localization data, and small molecule analog data. 
     
     
         47 . The method of  claim 45 , wherein the bioactivity profile comprises drug efficacy data and drug safety data. 
     
     
         48 . The method of  claim 47 , wherein the bioactivity profile further comprises at least one of gene expression data, mammalian LD50 data, metabolic data, physiological transport dynamics, pharmacokinetic data, water solvation data, absorption data, excretion or clearance data, liver toxicity data, membrane permeability data, cellular localization data, small molecule analog data, diffusion coefficient, and oral bioavailability data. 
     
     
         49 . The method of  claim 43 , wherein the biological sequences are selected from the group consisting of DNA nucleotide sequences, RNA nucleotide sequences, amino acid sequences, chemically modified DNA nucleotide sequences, biologically modified DNA nucleotide sequences, chemically modified RNA nucleotide sequences, biologically modified RNA nucleotide sequences, chemically modified amino acid sequences, biologically modified amino acid sequences, chemically modified protein sequences, and biologically modified protein sequences. 
     
     
         50 . The method of  claim 43 , wherein the desired bioactivity is an effect selected from the group consisting of biochemical, biophysical, pharmacological, therapeutic, antimicrobial, cytotoxic, antitumor, antiproliferative, and antineoplastic. 
     
     
         51 . The method of  claim 43 , wherein the distinguishing features of the validated biomolecules with the desired bioactivity profile are associated with a chemical physical, biological, pharmacological, or clinical functionality of an amino acid sequence, a RNA sequence, or a DNA sequence, constituting or coding for a molecule with anti-bacterial, anti-viral, anti-fungal, anti-parasitic, or anti-pathogen activity.