Use of protein in predicting drug properties

A method for predicting the toxicity of a drug, wherein the method comprising:

• (1) preparing a buffer solution:
• preparing a 0.02 M phosphate buffer solution with pH value of 7.4;
• (2) preparing a protein diluent:
• dissolving and diluting a protein solution with the buffer solution prepared in step (1) to obtain a protein diluent;
• (3) preparing a detection solution:
• mixing the protein diluent prepared in step (2) with a drug to be tested in a molar ratio of protein to drug of 1:
• (1-300) to obtain a mixed solution to be tested, and predicting the properties of the drug by
• using fluorescence emission spectrum, and wherein the protein comprises human serum albumin (HSA), bovine serum albumin, hemoglobin, globulin, myogenic, collagen, zymoprotein, bee protein and fish protein, the method for predicting the drug toxicity further comprises a step of preheating the mixed solution to be tested at 232-283 K for 3 minutes, and measuring fluorescence emission spectrum of the protein mixed with the drugs at 300-500 nm, an excitation wavelength is 280 nm, and a slit width for both excitation and emission are 15 nm;
• concentration of the protein is 5×10−7 M, and the concentration of the drug is (0-105)×10−7 M;
• correcting an internal filtering effect of fluorescence signal according to the following formula:
• Fcor=Fobsexp[(Aex+Aem)/2] (1);
• in this formula, Fcor and Fobs represent corrected fluorescence signal and observed fluorescence signal, respectively, and Aex and Aem represent absorbance of the mixed solution to be tested at excitation wavelength and emission wavelengths, respectively,
• wherein the drug is a pesticide selected from the group consisting of aldicarb, oxamyl, thiofanox sulfone, carbofuran, thiofanox, methomyl, aldicarb sulfone, aminocarb, promecarb, propoxur, dimethacarb, ethiofencarb, XMC (3,5-xylyl methylcarbamate), metolcarb, fenobucarb, isoprocarb and butoxycarboxim,
• wherein the method for predicting the drug toxicity further comprises a step of obtaining five values comprising a quenching constant, a binding constant, a number of binding sites, a binding distance and a free energy of the drug by fluorescence emission spectra, and then constructing a model using a backpropagation neural network, obtaining an LD50 value of the drug by inputting into the model the said five values detected by fluorescence spectrum of the drug, and obtaining a toxicity grade according to the criteria for pesticide hazard classification from the World Health Organization, thus completing a toxicity prediction.