1. Computational Metagenomics Lab
  2. Untitled project
  3. PanPhlAn


Clone wiki

PanPhlAn / Home

PanPhlAn - strain detection and characterization

Pangenome-based Phylogenomic Analysis

PanPhlAn is a strain-level metagenomic profiling tool for identifying the gene composition and in-vivo transcriptional activity of individual strains in metagenomic samples. PanPhlAn’s ability for strain-tracking and functional analysis of unknown pathogens makes it an efficient tool for culture-free infectious outbreak epidemiology and microbial population studies.


Based on shotgun metagenomic samples, PanPhlAn enables:

  • strain identification and characterization: of unknown strains in metagenomic samples. The gene set of strains present in samples is detected by screening for all potential genes from the species pangenome.
  • outbreak monitoring: pathogen detection and characterization, see E. coli example below.
  • population genomics: exploring the diversity of a species based on detected strains in hundreds of samples, see → Example of E. rectale and A. muciniphila below.
  • strain tracking: detecting identical gene content profiles of strains in different samples
  • functional analysis: based on detected strain-specific genes, the gene sequences can be used for functional investigations using KEGG
  • in-vivo transcriptional activity: with focus on genes that are specific to the individual strain in a sample (species and strain specific transcriptomics).


PanPhlAn is written in Python and runs under Ubuntu/Linux. It requires additional software tools to be installed on your system, including Bowtie2 and Samtools, see our wiki:
→ Download and Installation.

Contact & User support

→ Tutorial - PanPhlAn step by step
→ User forum - discussion group

The PanPhlAn software team: Matthias Scholz (algorithm design), Thomas Tolio (programmer), and Nicola Segata (principal investigator). For help, write to panphlan-users@googlegroups.com.

The PanPhlAn steps

  1. Create a pangenome database of a bacterial species, or download a pre-processed database
    ./panphlan_pangenome_generation.py -c speciesname --i_fna genome-files/ --i_ffn gene-files/ -o database/
    →
  2. Map each metagenomic sample against the species database.
    Example: screen for E. coli pangenome genes in sample01 and sample02 (ecoli16 database)
    ./panphlan_map.py -c ecoli16 -i Samples/sample01.fastq -o map_results/sample01_ecoli16.csv
    ./panphlan_map.py -c ecoli16 -i Samples/sample02.fastq -o map_results/sample02_ecoli16.csv
    →
  3. Merge and process the mapping results for getting the final gene-family presence/absence profile matrix
    ./panphlan_profile.py -c ecoli16 -i map_results/ --o_dna result_gene_presence_absence.csv
    →

Example of E. coli strain profiling

Characterization of the German 2011 E. coli outbreak strain

German 2011 E. coli outbreak

PanPhlAn profiling of the German outbreak metagenomes using a reference database in which the target outbreak genome is missing. (a) Hierarchical clustering. The heatmap displays presence/absence gene-family profiles of 110 reference strains (bright colored columns) and of 12 metagenomically detected strains (darker columns). Most outbreak samples cluster together due to almost identical profiles (right), with four samples (left) showing different profiles due to the presence of additional dominant E. coli strains overlying the target outbreak strain. (b) Functional analysis of outbreak-specific gene-families (Fisher exact test) confirmed that the outbreak strain is a combination of a EAEC pathogen (pAA plasmid) with acquired Shiga toxin and antibiotic resistance genes, complemented with a set of enriched virulence-related functions and pathway modules.

Example of Population Genomics Studies

Exploring the populations of Eubacterium rectale and Akkermansia muciniphila

PanPhlAn E. rectale A. muciniphila

Large-scale population genomics study of E. rectale and A. muciniphila. Based on 1830 metagenomic samples from 8 cohorts, PanPhlAn reveals the subspecies structure even when only few species reference genomes are available. (a) Based on only one reference genome, E. rectale strains can be resolved into three geographically distinct clades. Clade A is related to samples of the two Chinese cohorts (bright and dark green dots). (b) Based on two available reference genomes, PanPhlAn shows a clear cluster structure of A. muciniphila strains, suggesting that the species can be distinguished into six functionally distinct clades A-F.


Matthias Scholz*, Doyle V. Ward*, Edoardo Pasolli*, Thomas Tolio, Moreno Zolfo, Francesco Asnicar, Duy Tin Truong, Adrian Tett, Ardythe L. Morrow, and Nicola Segata (* Equal contribution)
Strain-level microbial epidemiology and population genomics from shotgun metagenomics
Nature Methods, 13, 435–438, 2016.

PanPhlAn is a project of the Computational Metagenomics Lab at CIBIO, University of Trento, Italy.