It's essential to consider the following possibilities:
Protonation and Tautomeric Forms: In some cases, AutoDock Vina may predict the most stable or favorable protonation or tautomeric form of your molecule during the docking process. This can lead to changes in the molecular structure to optimize binding interactions.
Force Field and Parameterization: AutoDock Vina employs force fields and energy minimization techniques to predict binding conformations. The force field and parameterization used in docking software may affect the molecular structure. Changes like bond rotations or adjustments to minimize energy can occur.
Molecule Flexibility: Some docking programs, including AutoDock Vina, allow for molecular flexibility during the docking process. Flexible ligand docking considers variations in bond angles and torsional rotations. As a result, the molecule may adapt to its predicted binding pose.
Incorrect Input: Check if the input file provided to AutoDock Vina in PyRx accurately represents the molecular structure you intend to dock. Incorrect atom types, charges, or connectivity could lead to changes during the docking.
Output Visualization: Ensure that the visualization software you use to view the results does not introduce any artifacts. Some visualization tools may not accurately depict the output structures.
To address these issues and potentially prevent undesired structural changes, you can consider the following steps:
Confirm the correctness of the input file and its format.
Experiment with different software settings, including constraints and flexibility options.
Consult the AutoDock Vina and PyRx documentation for specific options that may influence the output structure.
Validate the output structures using multiple visualization tools to ensure consistency.
Single bond vs double bond is most likely due to visualization error or rather your structure error. It normally happens when your bond are not fully satisfied, so the software thinks that there might be double bond here as with PDB and, I hope, with pdbqt, there is no bonding information in the file. Therefore, visualization software has to guess these bonds based on proximity and sometimes they are not very accurate.
I hope you have provided the minimized or clean inputs before docking. Also carefully check the ligand file when you convert them in PDBQT. I faced similar problem latter I found there was a problem during the ligands format conversion.