Biology Reference
In-Depth Information
residues lining the active site to narrow down its predictions. 37 This tool is available online
via NRPSpredictor2. 38 For the analysis and comparison of the nonribosomal peptide
products themselves, the database NORINE was established, which currently contains over
1100 compounds. 39
For the analysis of PKSs, the earliest computational tools focused on the identification of
their constituent domains 40,41 and of the linker regions between them. 42 The purpose here
was not only to aid product prediction, but also to facilitate the reconstitution of individual
domains and the combinatorial biosynthesis of
natural products via domain
swapping. Identification of the specificity-determining residues of acyltransferase domains
was again enabled by crystallographic data, allowing prediction of malonyl-CoA
or methylmalonyl-CoA specificity. 40,41 More recent tools have afforded more accurate
predictions and increased functionality. For example, ASMPKS allows the input of entire
genome sequences for analysis. 43 As with NRPSs, support vector machines have been
applied to type III PKSs to afford improved prediction accuracy. 44 Finally, SBSPKS allows
PKS domains and modules to be modeled and docked to better predict and engineer
intersubunit contacts, as well as to predict the order of substrate channeling in gene clusters
with multiple PKS open reading frames. 45
'
unnatural
'
Currently, the bulk of the state-of-the-art computational tools for analysis of secondary
metabolites focus on concomitant analysis of both PKS and NRPS genes. These tools,
such as NRPS-PKS, 46 ClustScan, 47 CLUSEAN, 48 NP.Searcher, 49 and the PKS-NRPS Analysis
Web-site 50 continue to improve upon the accuracy and functionality of their predecessors
while providing user-friendly interfaces and increased computational power. Nevertheless,
other classes of secondary metabolites have seen increased attention in recent years as well.
For example, the genome mining tool BAGEL2 focuses exclusively on bacteriocins,
which are ribosomally synthesized antimicrobial peptides from bacteria. 51 BAGEL2
considers conserved domains, physical properties, and genomic context to identify putative
bacteriocins, which can easily be missed by other genome annotation tools due to their
short size (
187
100 amino acids). Another example is the SMURF program, which can be used
to scan fungal genome sequences not only for gene clusters producing polyketides and
nonribosomal peptides, but also for indole alkaloids and terpenes by identifying
prenyltransferases and terpene cyclases, respectively. 52 Finally, a new program called
antiSMASH promises the greatest versatility of any program to date, expanding beyond
just polyketides and nonribosomal peptides to include such classes of compounds as
β
,
-lactams, lantibiotics, and siderophores, among others. 53 The identification and analysis of
such diverse secondary metabolism genes will surely be of great benefit to the synthetic
biologist for the design and engineering of new pathways. Of course, it must be stated that
despite their increasing utility, in silico tools will never completely replace laboratory
experiments for the analysis and understanding of secondary metabolite genes. Nevertheless,
such tools are clearly a very valuable component of the synthetic biology toolkit.
DE NOVO PATHWAY CONSTRUCTION AND OPTIMIZATION
While the identification and analysis of genes, proteins, and pathways found in nature
provide an excellent starting point, it is the goal of the synthetic biologist to go above and
beyond natural biological systems to obtain improved, or even completely novel,
functionalities. The former of these goals has been addressed by a number of in silico
experiments and tools designed to predict genetic modifications that will improve
productivity of a particular value-added compound. An early example of this was provided
by Lee and coworkers in 2002, who modeled the metabolism of E. coli to study succinic
acid synthesis and predict a more efficient pathway. 54 One of the landmark computational
tools for strain optimization is OptKnock, originally described by the Maranas group in
2003. 55 OptKnock utilizes a bilevel optimization framework to reshape the metabolic
network of E. coli such that the desired product is a necessary byproduct of growth.
Search WWH ::




Custom Search