Improving pairwise comparison of protein sequences with domain co-occurrence

Area of Science:

• Bioinformatics
• Computational Biology
• Genomics

Background:

• Protein sequence comparison is crucial for identifying conserved domains and motifs.
• Standard tools like BLAST can miss significant hits due to stringent thresholds, especially for distantly related species.
• Integrating contextual information can improve the sensitivity of sequence comparison.

Purpose of the Study:

• To develop a method that increases the sensitivity of pairwise protein sequence comparisons using domain co-occurrence.
• To construct new protein domain families based on co-occurrence data.
• To evaluate the method's effectiveness using Plasmodium falciparum.

Main Methods:

• Incorporating domain co-occurrence information into standard BLAST analysis.
• Developing a method to construct new domain families from co-occurrence results.
• Case study analysis using Plasmodium falciparum proteome data.

Main Results:

• Achieved a 14% increase in significant BLAST hits.
• Expanded proteome coverage by 25% with identified domains.
• Discovered 2240 new protein domains, many not linked to existing Pfam models.
• New domains exhibit alignment and physicochemical properties comparable to standard Pfam domains.

Conclusions:

• Domain co-occurrence is a valuable feature for enhancing protein sequence comparison sensitivity.
• The proposed method effectively identifies novel protein domains and expands proteome annotation.
• The newly identified domains are of comparable quality to established databases like Pfam.