1. Computational genomic strategies for natural product discovery
Dr. Marnix H. Medema
Bioinformatics Group
Wageningen University, The Netherlands
EBI Course Exploiting Metagenomics
Thursday, december 3rd, 2015, 11:00h

2. Microbial Biosynthetic pathways:
a great source of valuable molecules

3. Specialized metabolites play key roles in microbiomes

4. Specialized metabolites play key roles in microbiomes
Donia et al. (2014) Cell 158:

5. Diverse and complex enzymology produces chemical diversity: riPPs
Huge assembly-lines,
But: not the only mechanism
RiPPs: Ribosomally synthesized and Posttranslationally modified Peptides
Ortega et al. (2015), Nature 517:

6. nonproteinogenic amino
Diverse and complex enzymology produces chemical diversity: nonribosomal peptides
Key enzyme class: Nonribosomal Peptide Synthetase (NRPS)
NRPSs can introduce
nonproteinogenic amino
acids into peptides!
Huge assembly-lines,
But: not the only mechanism

7. Diverse and complex enzymology produces chemical diversity: nonribosomal peptides
Huge assembly-lines,
But: not the only mechanism
Schmartz et al. (2014), Nat. Prod. Rep. 12:

8. Diverse and complex enzymology produces chemical diversity: polyketides
Key enzyme class: Polyketide synthase (PKS)
Huge assembly-lines,
But: not the only mechanism
Menzella et al. (2005), Nat. Biotechnol. 23:

9. Diverse and complex enzymology produces chemical diversity: polyketides
Not all polyketide synthases are modular, some are iterative!
Fungal Type I
Type II
Type III
etc.
Huge assembly-lines,
But: not the only mechanism
Shen et al. (2003), Curr. Opin. Chem. Biol. 7:

10. Diverse and complex enzymology produces chemical diversity: terpenes
Huge assembly-lines,
But: not the only mechanism
Key enzyme classes: terpene synthases / cyclases
These turn isoprene precursors into mature terpenoids
Gao et al. (2012), Nat. Prod. Rep. 29:

11. Diverse and complex enzymology produces chemical diversity:
saccharides
Key enzyme class: glycosyl transferase
Huge assembly-lines,
But: not the only mechanism
McCranie & Bachmann et al. (2014), Nat. Prod. Rep. 31:

12. Biosynthetic gene clusters: the genetic basis of molecular diversity
So if we can find new gene clusters, we can find new chemicals! Now how to find new gene clusters?

13. Modularity of biosynthetic gene clusters
Second strategy
Cacho et al. (2015) Front. Microbiol 5: 774.

14. Modularity of biosynthetic gene clusters
Second strategy
Medema, Cimermancic et al. (2015) PLoS Comp. Biol. 10: e

15. antiSMASH: A Web Server for the Detection and analysis of biosynthetic gene clusters15
Medema et al. (2011) Nucl. Acids Res. 39: W339-W346.
Blin, Medema et al. (2013) Nucl. Acids Res. 41: W

16. Core structure prediction for polyketide synthase and nonribosomal peptide synthetase gene clusters16
Medema et al. (2011) Nucl. Acids Res. 39: W339-W346.
Blin, Medema et al. (2013) Nucl. Acids Res. 41: W

17. Comparative analysis and subcluster detection17
Medema et al. (2011) Nucl. Acids Res. 39: W339-W346.
Blin, Medema et al. (2013) Nucl. Acids Res. 41: W

18. Another Method to Detect Biosynthetic Gene Clusters in Prokaryotic Genomes
Training set consisted of 732 biosynthetic gene clusters of known compounds:
136 type I polyketides
100 nonribosomal peptides
76 type II polyketides
82 polyketide-peptide hybrids
93 oligo- and polysaccharides
38 aminoglycosides
36 terpenoids
27 ribosomal peptides
23 lantibiotics
13 indolocarbazoles
11 type III polyketides
9 fatty acids
9 siderophores
8 nucleosides
6 beta-lactams
4 aminocoumarins
61 others
Cimermancic, Medema, Claesen et al. (2014) Cell 158:

19. Large metagenomic datasets may contain very large numbers of biosynthetic gene clusters
Now there are of course both rare and frequently occurring classes of gene clusters / compounds. What we had not expected was to find large clusters within this network that contain no known gene clusters.
We chose one of these regions, which contained two related families of hundreds of gene clusters encoding amongst others very unusual ketosynthases CoA-ligases.
Cimermancic, Medema, Claesen et al. (2014) Cell 158:

20. Data on bgcs is scattered and not systematically stored

21. The minimum information about a biosynthetic gene cluster (MIBiG)21
Medema et al. (2015) Nature Chem. Biol., under review.

22. a rich set of annotations and metadata on biosynthetic gene clusters
General MIBiG Parameters
Biosynthetic class
MIxS environmental / taxonomic information
Number of loci
Complete / partial cluster
Nucleotide sequence accession
16S accession / sequence
Custom gene names
Functional sub-clusters
Biosynthetic genes
Transport-related genes
Regulatory genes
Resistance/immunity genes
Operon architecture
Knockout mutant phenotypes
Compound name
Synonyms for compound name
Exact molecular mass
Molecular formulae of the compound(s)
Compound structure
Chemical moieties
Compound activity
Compound molecular target
Publications on activity/toxicity/target
Tailoring reactions
Evidence for compound-cluster connection
Polyketide-specific
Polyketide synthase type
Polyketide subclass
Linear / cyclic
PKS genes
Number of PKS modules
Ketide unit sequence
Starter Unit
Reductive domains
KR stereochemistries
AT domain substrate specificities
Non-reductive modifying PKS domains
Module skipping / iteration
Number of iterations (if iterative)
Iterative PKS subtype (if iterative)
Trans-acyltransferase genes
Inactive / atypical domains
TE domain type
Cyclization / termination type
Nonribosomal peptide-specific
NRP subclass
Linear / cyclic
NRPS genes
Number of NRPS modules
NRP amino acid sequence
A domain substrate specificities
Variable A domain specificities
Condensation domain subtypes
Modifying domains (Me/Ox/Red/Epi)
Module skipping / iteration
TE domain type
Cyclization / termination type
RiPP-specific
RiPP subclass
Linear/cyclic
Precursor-encoding gene(s)
Precursor peptide length
Leader peptide length
Follower peptide length
Core peptide length
Core peptide sequence
Cleavage recognition site
Number of crosslinks
Crosslink positions
Type of crosslinks/cyclizations
Recognition motif in leader peptide
Terpenoid-specific
Terpene subclass
Precursor carbon chain length
Final isoprenoid precursor
Terpene synthases / cyclases
Prenyltransferases
Saccharide-specific
Saccharide subclass
Glycosyltransferase (GT) genes
GT substrate specificities
Alkaloid-specific
Alkaloid subclass
Specific for other classes
Biosynthetic class specification 22
Again, MIBiG has an important role to play here, as standardized data submission and storage will allow us to build up a parts registry that can function as a trustworthy repository for designing new pathways.
Medema et al. (2015) Nature Chem. Biol., under review.

23. >75 research groups worldwide participated
Community annotation of biosynthetic gene clusters using MIBiG 23
>75 research groups worldwide participated
Result: detailed annotation of ±400 BGCs, essential annotations for another ±900 BGCs
So we currently have a draft version of MIBiG, on which between PIs in the field have already commented through an online survey.
Later this week, I will organize a discussion session, to which I would like to invite you all to discuss this further.
A standard has to be carried by the community.

24. Community annotation of biosynthetic gene clusters using MIBiG24
So we currently have a draft version of MIBiG, on which between PIs in the field have already commented through an online survey.
Later this week, I will organize a discussion session, to which I would like to invite you all to discuss this further.
A standard has to be carried by the community.

25. An online repository for MIBIG information25

26. Integration with antismash: KnownClusterblast26

27. Finding more variants of known enzymatic parts using Multigeneblast27
Medema et al. (2013) Mol. Biol. Evol. 30:

28. Finally: some suggestions for analyzing metagenomes using antismash28
Assemble first!
Only run contigs > 2 kb; use other tools for very fragmented assemblies, e.g.
Sort contigs by size, if >1000 contigs: run locally or contact us to run it on the public server
Local installations: Docker container available