Functional (re)annotation of Mycobacteroides abscessus proteome using integrative sequence and AI-based structural...

Area of Science:

• Proteomics and Bioinformatics
• Microbiology and Infectious Diseases

Background:

• Functional annotation of proteins is essential for understanding organism biology and pathogenesis, particularly for opportunistic pathogens like Mycobacteroides abscessus (MAB).
• A significant portion of the MAB proteome remains poorly annotated, hindering the understanding of its functional landscape, drug resistance mechanisms, and potential for transmission.
• There is a critical need to improve the functional descriptions and Gene Ontology (GO) terms for MAB proteins to identify potential drug targets.

Purpose of the Study:

• To systematically functionally (re)annotate the Mycobacteroides abscessus proteome using a combination of sequence and AI-driven structure-based approaches.
• To address the substantial gap in functional annotations for MAB proteins, including essential genes.
• To leverage predicted protein structures and sequence similarity for improved functional characterization.

Main Methods:

• Performed sequence-based similarity searches against the NR database and HMM-based searches for functional domains (Pfam, CATH).
• Utilized AI-predicted structures (AlphaFold) for MAB proteins and performed structure-based similarity searches (Foldseek) to transfer GO annotations.
• Integrated sequence and structure-based approaches for proteins lacking AlphaFold structures.

Main Results:

• Assigned new GO annotations to 374 MAB proteins and refined existing annotations for 885 proteins, including previously unannotated essential genes.
• Successfully applied a combined sequence- and structure-based annotation strategy, even for proteins without available AlphaFold structures.
• Identified residue-level differences in MAB proteins homologous to Mycobacterium tuberculosis, potentially linked to drug resistance.

Conclusions:

• The study demonstrates the efficacy of a combined sequence- and AI-driven structure-based approach for large-scale proteome functional annotation.
• This methodology significantly enhances the functional understanding of the Mycobacteroides abscessus proteome, aiding in the study of pathogenesis and drug resistance.
• The developed approach is broadly applicable to functional proteome annotation in other organisms of interest.