mdCATH: A Large-Scale MD Dataset for Data-Driven Computational Biophysics.

Area of Science:

• Structural Biology
• Computational Biology
• Biophysics

Background:

• Protein structure determination has advanced, but data on protein dynamics is scarce.
• Protein dynamics are essential for understanding protein function, folding, and interactions.
• Existing datasets lack comprehensive coverage of protein domain dynamics.

Purpose of the Study:

• To address the gap in protein dynamics data by creating a large-scale dataset.
• To provide a resource for proteome-wide statistical analyses of protein dynamics.
• To facilitate research into protein unfolding thermodynamics and kinetics.

Main Methods:

• Generated mdCATH dataset using extensive all-atom molecular dynamics simulations.
• Simulated 5,398 protein domains across five replicates and five temperatures (320 K to 450 K).
• Recorded coordinates and forces every 1 ns, accumulating over 62 ms of simulation time.

Main Results:

• Created mdCATH, a comprehensive dataset capturing protein domain dynamics.
• The dataset includes detailed simulation data for a diverse collection of protein domains.
• Demonstrated the dataset's utility through reproducible case studies.

Conclusions:

• mdCATH provides a unique resource for advancing protein science.
• The dataset enables deeper insights into protein folding, function, and interactions.
• Facilitates statistical analyses of protein unfolding thermodynamics and kinetics.