PathogenFinder

PathogenFinder predicts the pathogenic potential of bacterial strains for human hosts from proteomes, whole genomes, or raw sequencing reads.

Key Features:

• Data Input Flexibility: Accepts proteomes, whole genomes, or raw sequencing reads as input.
• Taxonomic Universality: Functions across all taxonomic groups of bacteria.
• Non-Biased Methodology: Analyzes groups of proteins without considering annotated functions or known pathogenicity-associated genes.
• High Predictive Accuracy: Achieved 88.6% accuracy on an independent test set, correctly classifying 398 of 449 completely sequenced bacteria.

Scientific Applications:

• Pathogenicity Prediction: Predicts the pathogenic potential of newly sequenced bacterial strains to assess potential impact on human health.
• Discovery of Novel Virulence Factors: Aids identification of previously unrecognized virulence factors by not relying solely on known pathogenic genes.
• Strain Characterization: Isolates potential pathogenic features in well-characterized and novel bacterial strains to support targeted biological investigations.

Methodology:

Protein groups are created from input proteomes, genomes, or reads and analyzed to predict pathogenicity, with the analysis based on a comprehensive training set comprising pathogenic and non-pathogenic bacterial genomes.