Mining recombination algorithms in modular biosynthetic gene clusters

Non-ribosomal peptide synthetases (NPRS) are ubiquitous in micro-organisms and produce large varieties of metabolites with pharmaceutical potentials. Its modularity structure inspired de-novo designing efforts by recombining desired modules. However, past efforts in this direction are largely unsuccessful. The difficulty in re-engineering new products raise questions on the modulized view: are there unknown functional constraints between different parts of NRPS that favor/disfavor certain combinations of modules or subunits?

In large database, correlations between different subunits of NRPS can be uncovered by statistic techniques. However, as subunits in the same gene cluster are correlated by sharing the same lineage history, the correlation induced by functional constraints usually get blurred.

A unique marine bacteria described in our previous research reveals important design principles of NRPS. In this bacteria, 20 NRPS pathways generate chemical diversity though extremely frequent recombination. Genetic exchange between NRPSs is so intensive that it largely erases the correlation by lineage history, leaving correlation by function clearly observable. In this system, we revealed previously undescribed constraints between different types of subunits of NRPS.