Cd-hit: a fast program for clustering and comparing large sets of protein or nucleotide sequences

Area of Science:

• Bioinformatics
• Computational Biology
• Genomics
• Proteomics

Background:

• The cd-hit program, introduced in 2001-2002, provides ultrafast protein sequence clustering for large databases.
• The underlying algorithm's efficiency extends beyond protein clustering applications.
• Existing sequence comparison tools like BLAST can be computationally intensive for massive datasets.

Purpose of the Study:

• To introduce novel sequence analysis tools based on the cd-hit algorithm.
• To extend the utility of the cd-hit algorithm to comparative analysis of two datasets and nucleotide sequences.
• To provide highly efficient solutions for clustering and comparing large-scale biological sequence data.

Main Methods:

• Development of new programs: cd-hit-2d for comparing two protein datasets, cd-hit-est for nucleotide sequence clustering, and cd-hit-est-2d for comparing two nucleotide datasets.
• Leveraging the established ultrafast algorithm from the original cd-hit program.
• Designed to handle datasets containing millions of sequences.

Main Results:

• The new cd-hit suite (cd-hit-2d, cd-hit-est, cd-hit-est-2d) demonstrates high efficiency in handling large protein and nucleotide sequence datasets.
• These tools offer significant speed advantages, being hundreds of times faster than traditional methods like BLAST for similar tasks.
• The programs successfully perform comparative analysis between two datasets and clustering of large sequence databases.

Conclusions:

• The cd-hit algorithm's underlying principles can be effectively applied to develop versatile and high-performance tools for biological sequence analysis.
• The newly developed cd-hit programs offer substantial computational savings for researchers working with massive sequence data.
• These tools represent a significant advancement in efficient database searching and sequence comparison within bioinformatics.