A fast and symmetric DUST implementation to mask low-complexity DNA sequences

Area of Science:

• Bioinformatics
• Computational Biology
• Genomics

Background:

• The DUST algorithm is a long-standing tool for masking low-complexity DNA sequences within the BLAST search suite.
• Existing masking rules present limitations, including lack of symmetry and context sensitivity.

Purpose of the Study:

• To introduce a novel implementation of the DUST algorithm with an enhanced masking strategy.
• To address deficiencies in the previous DUST masking approach, improving sequence analysis accuracy.

Main Methods:

• Developed a new DUST module implementation utilizing the established complexity scoring function.
• Introduced a revised masking rule that ensures symmetry and removes context dependency.
• Evaluated the new implementation's performance and impact on MegaBLAST outputs.

Main Results:

• The new DUST masking rule is symmetric and not context-sensitive, correcting prior limitations.
• The revised implementation demonstrates a significant speed improvement, being at least four times faster on the human genome.
• The additional masked bases and impact on MegaBLAST results were found to be minimal.

Conclusions:

• The enhanced DUST module offers a more robust and efficient method for low-complexity sequence masking in bioinformatics.
• This update improves sequence analysis by providing more reliable masking without compromising BLAST performance or introducing significant changes to results.