Improving the quality of twilight-zone alignments

Area of Science:

• Bioinformatics
• Computational Biology
• Structural Bioinformatics

Background:

• Recognizing distant homologies between proteins is crucial for understanding protein function and evolution.
• The accuracy of protein alignment methods directly impacts the reliability of homology detection.

Purpose of the Study:

• To evaluate the impact of evolutionary information in sequence profiles on protein alignment accuracy.
• To compare the performance of different alignment methods (sequence-sequence, sequence-profile, profile-profile) for distant homology recognition.

Main Methods:

• Utilized a database of protein pairs with similar structures but low sequence similarity.
• Assessed alignment quality across various methods, including sequence-profile and profile-profile alignment algorithms.
• Developed methods to standardize alignment lengths for fair comparison.

Main Results:

• Sequence profile incorporation significantly enhances alignment accuracy, approaching structure-based alignment quality.
• Alignment quality strongly correlates with statistically significant alignment scores and correct structural templates.
• Profile-profile methods, like FFAS, generally produce longer alignments that often include structurally relevant fragments.

Conclusions:

• Evolutionary information encoded in sequence profiles is vital for accurate protein alignment.
• Statistically significant alignment scores serve as reliable indicators of both correct template identification and high-quality alignments.
• Standardizing alignment lengths is necessary for robust comparisons between different algorithmic approaches.