Universal seeds for cDNA-to-genome comparison.

Area of Science:

• Genomics
• Bioinformatics
• Computational Biology

Background:

• Gene annotation requires accurate cross-species sequence alignment.
• Optimized spaced seeds aid alignment but have been tested mainly on closely related species or simulated data.
• A need exists for universal seeds applicable across diverse species comparisons.

Purpose of the Study:

• To investigate seed performance across different evolutionary distances in cDNA-to-genome comparisons.
• To identify universal seeds that perform optimally or near-optimally for broad applicability.
• To enable user-friendly gene annotation for newly sequenced organisms.

Main Methods:

• Statistical regression methods were used to analyze seed sensitivity.
• Four cDNA-to-genome comparisons at varying evolutionary distances were evaluated (human-dog, human-mouse, human-chicken, human-zebrafish).
• Seed behavior across different comparison classes was identified.

Main Results:

• Classes of comparisons exhibiting similar seed behavior were identified, allowing for shared seed usage.
• Good seeds optimized for distant comparisons performed well (98-99% of optimal) on closer comparisons.
• These seeds represent effective universal seeds for cross-species alignment.

Conclusions:

• Optimal seeds for distant species comparisons are broadly applicable across a wide range of comparisons.
• This finding facilitates the development of practical cDNA-to-genome alignment tools.
• The research aids in the annotation of new model organisms by providing universal seeds.