PELICAN Tool Improves Gene Recall in Bacteriophage Genome Annotation

Researchers have introduced PELICAN, a command-line tool for annotating bacteriophage genomes, in an open-access paper published in BMC Bioinformatics. Bacteriophages, the viruses that infect bacteria, are increasingly studied as antibacterial agents against important bacterial pathogens, and accurate gene identification in their genomes is essential for both basic research and therapeutic applications. Detecting genes linked to lytic or lysogenic lifestyles, toxins, or bacterial fitness helps guide the selection of phages for such uses.

The PELICAN pipeline operates in four stages. First, it integrates predictions from multiple annotation tools. Second, it applies probabilistic and machine learning-based evaluation to non-consensus genes, the predictions that different tools do not agree on.

Third, it resolves overlapping predictions, situations where candidate coding regions occupy the same genomic positions. Finally, it performs a codon bias check of overlapped positions to produce a final annotation. To validate the approach, the team used reference phage genome annotations that had been validated by transcriptomics, an experimental method that provides evidence of which regions are actually expressed.

PELICAN's results were then compared with those of other phage genome annotation tools. In these comparisons, PELICAN achieved the best recall, meaning it detected the largest share of the genes present in the reference annotations. However, this sensitivity came with a higher false positive rate than some of the competing tools produced.

The authors went on to examine whether those apparent false positives were genuinely errors. They present evidence that most of them may in fact be true positives. Their argument rests primarily on protein folding scores of the putative proteins encoded by the predicted coding regions, which suggest that many of the disputed predictions correspond to sequences capable of forming coherent protein structures.

A tool that flags many extra genes is not necessarily wrong if those genes are real but overlooked by the reference annotations, and the folding-score evidence offers a line of support for that interpretation. Taken together, the validation results support the paper's conclusion that PELICAN provides improved gene recall and competitive F1 values compared to an existing similar tool. PELICAN is presented as a novel option for bacteriophage genome annotation, aimed at researchers who need to characterize phage genomes for study or for therapeutic development.

The work was carried out by Fernando P. N. Rossi, Guillermo Uceda-Campos, Guilherme Wenceslau de Lima Cardoso, Pedro Rossi de Andrade Franco and João Carlos Setubal.

Subject of Research: PELICAN: a new bacteriophage genome annotation tool

Article Title: PELICAN: a new bacteriophage genome annotation tool

Article References: Rossi, F. P. N., Uceda-Campos, G., Cardoso, G. W. D. L., Franco, P. R. D. A., & Setubal, J. C. (2026). PELICAN: a new bacteriophage genome annotation tool. BMC Bioinformatics. https://doi.org/10.1186/s12859-026-06703-w

Image Credits: AI Generated

DOI: 10.1186/s12859-026-06703-w

Keywords: PELICAN, bacteriophage, genome, annotation, tool, scientific research