Model training method based on affinity data of biomolecule pair and related device

The embodiment of the invention discloses a model training method based on affinity data of a biomolecule pair and a related device. The method comprises the following steps: acquiring a target biological experiment group; performing affinity relationship prediction on the first biomolecule pair and the second biomolecule pair by using an affinity relationship prediction model to obtain a predicted affinity relationship between the first biomolecule pair and the second biomolecule pair; pre-training the affinity relationship prediction model based on the predicted affinity relationship between the first biomolecule pair and the second biomolecule pair and the affinity tag of each biomolecule pair; migrating the pre-trained affinity relationship prediction model into the affinity data prediction model; and performing model fine tuning on the affinity data prediction model to obtain a fine-tuned affinity data prediction model. According to the embodiment of the invention, the affinity prediction performance of