"Docking score is terrible, but in experiments it's a potent inhibitor."
Akira was perplexed.
"Contradictory," Lina peered at the screen.
"What's wrong?"
"Could the protein structure be incorrect?"
"It's a crystal structure, so it should be accurate..."
Lina noticed. "Wait. Is this an apo structure?"
"Apo?"
"The protein without ligand bound."
"Yes. So I docked the compound here."
"That's the trap."
"Trap?"
Lina explained. "Proteins are flexible. When ligands bind, the structure changes."
"Induced-Fit... induced fit," Sena entered and said.
"Yes. Like a keyhole changing shape to fit the key."
Akira understood. "So docking to the apo form doesn't work."
"Exactly. The apo pocket is closed or has a different shape."
Sena asked. "So what should we do?"
"Use the holo structure. The ligand-bound state."
"But there's no holo structure of this compound."
"That's the problem," Lina nodded.
Akira pondered. "Then how do we account for Induced-Fit?"
"There are several methods."
Lina began listing.
"First, flexible docking. Make part of the protein flexible and adjust structure during docking."
"Part? Not all?"
"Making all atoms flexible causes computational cost to explode. So only around the binding pocket."
"I see."
"Second, ensemble docking. Prepare multiple structures and dock to each."
"Multiple structures?"
"Generate with molecular dynamics simulations. Move the protein to sample various conformations."
Sena was interested. "Can you see proteins moving?"
"Yes. MD simulation. Simulating time evolution."
Akira had another question. "Third?"
"Use holo forms of similar ligands. If there's a structure with a similar compound bound, use it as reference."
"What if that doesn't exist either?"
Lina sighed. "That's the most difficult."
"What do you do?"
"Homology modeling. Predict the structure of your target protein from similar proteins' holo forms."
"Uncertainty increases," Akira said.
"Yes. So ultimately experimental validation is necessary."
Sena hit the core. "When there's Induced-Fit, computational drug design is difficult."
"Difficult, but not impossible," Lina said.
"For example?"
"HIV-1 protease. A famous Induced-Fit example."
Lina displayed the structure. Apo with open flaps and holo with closed flaps.
"They differ this much?" Sena was surprised.
"Yes. But MD simulation can reproduce this change."
"Amazing."
Akira asked. "Computation time?"
"Hours to days. Depends on resources."
"Whether it's practical is questionable."
"So we choose based on application. Rigid docking for initial screening, MD and FEP for precise evaluation of final candidates."
Sena summarized. "Induced-Fit is a trap, but can be addressed with appropriate methods."
"Yes. If you know the trap, you can deal with it," Lina smiled.
Akira started new docking. "This time I'll use a structure generated by MD."
Calculation ran. After a few minutes, results came.
"Score improved!"
"Because you used a structure adapted to the ligand."
Sena was impressed. "Considering that proteins move is important."
"Life is dynamic. You can't see the truth looking only at static structures," Lina said.
"But," Akira added, "balance with computational cost is also necessary."
"Yes. Seeking perfection takes too much time."
Sena murmured. "The trap and salvation of Induced-Fit. Understanding that is the key to drug discovery."
Lina nodded. "Master flexibility, master docking."
The phenomenon called Induced-Fit. It's both a trap and an opportunity.