Combining local-structure, fold-recognition, and new fold methods for protein structure prediction.

Area of Science:

• Computational Biology
• Structural Bioinformatics
• Protein Structure Prediction

Background:

• Accurate protein fold recognition and de novo structure prediction are crucial for understanding protein function.
• Existing methods often face challenges in handling diverse protein structures and achieving high prediction accuracy.

Purpose of the Study:

• To present the SAM-T02 method for automated protein fold recognition.
• To introduce the UNDERTAKER program for ab initio protein structure prediction.
• To demonstrate the synergistic performance of combining SAM-T02 and UNDERTAKER.

Main Methods:

• The SAM-T02 method utilizes two-track hidden Markov models (HMMs) with amino acid and local structure alphabets for template protein alignment.
• The UNDERTAKER program employs fragment-packing algorithms, leveraging libraries generated from SAM-T02 alignments for de novo modeling.
• A combined approach integrates HMM-based fold recognition with fragment-based structure generation.

Main Results:

• The SAM-T02 server successfully identifies and aligns template proteins from the Protein Data Bank (PDB).
• The UNDERTAKER program generates protein conformations using fragment and alignment libraries derived from SAM-T02.
• The integrated method demonstrated high performance on selected test targets, including T0129, T0181, T0135, T0130, and T0139.

Conclusions:

• The combination of SAM-T02's fold recognition and UNDERTAKER's ab initio prediction offers a powerful approach for protein structure modeling.
• This integrated strategy enhances the accuracy and reliability of predicting protein folds and conformations.