I am using Diffdock-PP with different configurations and have noticed that by default, during docking, the first 5-6 poses (returned in visualization) cover different parts of the receptor. However, the subsequent poses focus on only one location, attempting to achieve the best fit. While Diffdock should correctly identify the target site when the protein is small, it sometimes fails to cover enough surface area in the initial steps when dealing with larger proteins, potentially missing the correct pose.
Is there a way to increase the number of initial steps to cover more surface area, or is this process dependent on the overall number of steps during docking?
I am using Diffdock-PP with different configurations and have noticed that by default, during docking, the first 5-6 poses (returned in visualization) cover different parts of the receptor. However, the subsequent poses focus on only one location, attempting to achieve the best fit. While Diffdock should correctly identify the target site when the protein is small, it sometimes fails to cover enough surface area in the initial steps when dealing with larger proteins, potentially missing the correct pose.
Is there a way to increase the number of initial steps to cover more surface area, or is this process dependent on the overall number of steps during docking?