The application discloses a kind of based on nanomechanics
fingerprint and one-dimensional
convolutional neural network (
deep learning) antiviral
agglutinin rapid
screening method,
system and storage medium, belong to the cross technical field of
biophysics,
nanotechnology and
artificial intelligence.This method realizes the in-situ capture of target
virus and
agglutinin binding reaction by constructing rigid seamless
liquid phase atomic force microscopic measurement interface;Atomic force
microscope is used to synchronously acquire the topographic image and force-
distance curve of
virus particle, and form multimodal nanometer characterization dataset.Extract the height, volume, roughness and other topographic features of
virus by
image processing algorithm, and extract the deep features of force spectrum
signal combined with
deep learning, input the multi-class
feature vector into the prediction model after fusion, calculate the inhibition
score of
agglutinin and complete the activity sorting, so as to realize the rapid screening of high-activity candidate agglutinin.The application not only solves the interference of substrate
instability and micro-bubble in
liquid phase AFM measurement, but also significantly improves the efficiency and accuracy of
antiviral drug screening, with good adaptability and popularization value.