The project was conducted following three main Aims: i) we developed and implemented an integrated assembly-based computational pipeline to reconstruct genomes of unknown organisms from large-scale metagenomic datasets. This comprised also the screening and curation of around 10,000 metagenomes coming from publicly available repositories and their integration with new cohorts acquired in the host laboratory. This effort resulted also in the development of the curatedMetagenomicData package (Pasolli et al., Nature Methods, 2017); ii) the panel of extracted genomes was characterized phylogenetically and taxonomically. This was done through an improved version of the PhyloPhlAn software able to place the reconstructed genomes within the known microbial tree of life and to infer their phylogenetic relationship; iii) we started to integrate the large set of reconstructed genomes with the already available reference genomes. This still ongoing task aims to improve microbiome-based host condition prediction approaches and strain-level profiling methods. For the latter, we also developed the novel metagenomic strain identification tool StrainPhlAn (Truong et al., Genome Research, 2017).
The developed pipeline was applied in multiple projects to answer to different biological relevant questions: i) we reconstructed bacterial and Eukaryotic genomes from a newly acquired cohort spanning 97 skin microbiomes associated with patients affected by psoriasis. We showed strain heterogeneity colonisation and functional variability providing the hypothesis of psoriatic niche-specic strain adaptation or selection (Tett, Pasolli, et al., npj Biofilms and Microbiomes, 2017); ii) we reconstructed genomes from a newly acquired cohort of mothers and infants from multiple body sites for a total of 216 metagenomes, providing evidence of vertical strain transmission at birth (Ferretti, Pasolli, et al., Cell Host & Microbe, 2018); iii) we conducted a large-scale metagenomics analysis from 1689 subjects for the characterization of Blastocystis in the human gut microbiome (Beghini*, Pasolli*, et al., ISME Journal, 2017); iv) we conducted a whole-genome epidemiology, characterisation, and phylogenetic reconstruction of 184 newly acquired Staphylococcus aureus strains in a paediatric hospital in Italy (Manara*, Pasolli*, Dolce* et al., Genome Medicine, 2018); v) we conducted a very large-scale analysis spanning body sites, ages, countries, and lifestyles. We reconstructed and characterized 154,723 genomes from a total of 9,428 metagenomes from 46 publicly available datasets in addition to a newly sequenced cohort from Madagascar. This resulted in the largest microbial catalogue of the human microbiome. We recapitulated 4,930 species-level genome bins (SGBs), 77% without genomes in public repositories, and annotated 2.85 M genes in SGBs, many associated with conditions including infant development or Westernization (Pasolli et al., Cell, 2019); vi) we extended the approach to non-human environments (Pinto et al., Genome Ann., 2017; Pinto et al., Genome Ann., 2018).
The objectives of the MSCA action have been fulfilled and exceeded from different points of view: i) the researcher Dr. Edoardo Pasolli has co-authored ten papers (in addition to other papers under review/preparation) in international journals, including four papers as (co-)first author in top journals (Cell, Nature Methods, ISME Journal, and Genome Medicine); 2) the researcher has successfully transitioned from developing machine learning methodologies to a more computational biology-focused research; 3) the researcher has strengthen his expertise in dissemination and communication capabilities thanks to his participation at multiple conferences and events; 4) the researcher has successfully transitioned from a postdoctoral to an independent position (as tenure-track assistant professor in Italy) in the EU system.
