SPECIES AT THE GENOMIC LEVEL

DDH has been the gold standard  the “sex” for higher eukaryotes Stackebrandt et al., 2002, Int J Syst Evol Microbiol. 52: Rosselló-Mora & Amann 2001, FEMS Rev. 25:39-67 Gevers et al., 2005, Nature Rev. Microbiol. 3: DDH (DNA-DNA hybridization):  70% similarity (50-70%)  used since the 60’s  strong influence  non cumulative DB  need to be substituted MLSA (multilocus sequence analysis):  5-10 full/partial sequences  house keeping genes  primer design difficulties  biases in the selection of genes  time consuming  ↓↓ number for stable topology Amplify and sequence 5-10 housekeeping genes for each strain Concatenate gene sequences Reconstruct the phylogeny genAgenBgenCgenDgenEgenF Str. 1 Str. 2 Str. 3 Str. 4

Alternative approaches  ANI Konstantinidis and Tiedje, 2005, PNAS. 102: Genome a BLAST N Genome b Search annotated ORFs sieve common orthologous genes ANI aa b genome Cut into fragments of 1020 nuc + BLAST N < 30% identity < 70% aligned seq > 30% identity > 70% aligned seq discard ANI Goris et al., 2007, IJSEM. 7:81-91

JSpecies ( JSpecies  Biologist oriented  user friendly and usable with multifasta data

ANI is way to circumscribe species genomically in the future ANIm vs DDH:  85 genospecies evaluated  94-96% a plausible borderline  inconsistent results most probably due to wrong DDH values  ANI thresholds of 94-96%  genomospecies  20% random sequences (i.e., 250 nuc) of two genomes is enough  Complete catalogue of type strain genomes  only 4% random genome sequence is enough Richter & Rosselló-Móra 2009, PNAS 106:

The best scenario ◄► all species genomes sequenced afedcb glkjih mrqpon sxwvut complete type strain genomes + < 20% random sequence genome coverage Perhaps with 1000 reads would be enough (200€) STABLE ANI  1% of the genome will be enough for IDENTIFICATION purposes  need of an effort to full sequence the species collection  need of an effort to full sequence the species collection (GEBA; Wu et al Nature 24: )  it will be in the future necessary to fully sequence any new type strain  94% - 96% ANI boundary

► Data analysis in summer 2009 => 938 genomes ► 10% of the entries tagged with the collection number (the rest with original strain number) ► 255 species names represented by their Type Strain ► 256 species names NOT represented by their Type Strain ► 50 species names NEVER validly published ► it is possible to circumscribe uncultured species (i.e. Buchnera & Wolbachia) Richter & Rosselló-Móra 2009, PNAS 106: Genome database & Type strains

Tetranucleotide variation: 4 4 = 256 TETRA:  Genomes have an oligonucleotide usage (not yet understood, related to codon usage)  Similar genomes might have similar usage  ALIGNMENT FREE PARAMETER  may be useful in deciding whether a group of strains deserve a species status  Same species >0.999

► The case of the synthetic genome of M. mycoides strain GM12 transplanted to M. capricolum (Science (2010) 329: 52) ► 88.5 (66% aligned) ► 94.5 (78% aligned) ► 87.8 (76% aligned)

► Only one of the several transplantations worked out! ► Different ways of reading the genome? organismtargetANI TETRA (r) M. hyopneumoniae 7448 M. hyopneumoniae J M. mycoides LC M. capricolum M. genitalium M. capricolum M. genitalium M. pneumoniae M. genitalium M. gallisepticum M. aligatoris (crocodyli) M. capricolum Same species WorkedNONONO Genome transplantation experiments of Venter

► The phylogenetic (evolutive) distance plays an important role in the recognition of how the genetic information is coded ► M. genitalium  M. pneumoniae, strange! Wrong identified strain?

OTHER PARAMETERS Average Aminoacid Identity (AAI) Kostantinidis & Tiejde, 2005, J. Bacteriol. 187: Maximal Unique and Exact Matches (MUM) De Loger et al., 2009, J. Bacteriol. 191: High Scoring Segment Pairs (HSP) (HSP) Auch et al., Std Gen Sci 2: And more to come Need full genome sequences The easiest is the best