The SAMPL2 blind prediction challenge: introduction and overview

Area of Science:

• Computational chemistry
• Physical chemistry
• Molecular modeling

Background:

• Molecular interactions with aqueous environments are crucial for chemical processes.
• Accurately predicting transfer energies and tautomer ratios in solution is computationally challenging.
• The SAMPL (Statistical Assessment of the Modeling of Proteins and Ligands) challenge aims to evaluate computational methods.

Purpose of the Study:

• To blind-assess the accuracy of computational methods for predicting transfer energies and tautomer ratios.
• To identify strengths and weaknesses in current computational approaches for molecular interactions in solution.
• To provide a benchmark for future method development in computational chemistry.

Main Methods:

• The SAMPL2 (Statistical Assessment of the Modeling of Proteins and Ligands) challenge involved blind prediction of transfer energies and tautomer ratios.
• Over 60 prediction sets were submitted by participants.
• Participants also attempted to estimate the error in their predictions.

Main Results:

• The blind assessment revealed limitations in current computational methods for predicting transfer energies.
• Predicting tautomer ratios in solution also proved challenging for most participants.
• Estimating prediction errors was difficult for the majority of participants, indicating a need for better uncertainty quantification.

Conclusions:

• Current computational methods have significant room for improvement in predicting molecular transfer energies and tautomer ratios in aqueous solutions.
• The SAMPL2 results highlight specific areas where computational chemistry methods are underperforming.
• Further development is needed to enhance the accuracy and reliability of computational predictions for molecular interactions in solution.