Biology Reference
In-Depth Information
to grow on minimal medium supplemented with these compounds, we considered the
assumption to be justifi ed.
Model-driven Reannotation
The reconstruction process systematizes knowledge about the metabolism of an or-
ganism, allowing the identification of errors in, and discrepancies between, various
sources of data. A major value of a manual model building effort is the careful revision
of the current genome annotation, based on literature evidence encountered during the
model building process, BLAST searches, and gap closures. During the reconstruc-
tion of the
P. putida
metabolic network, we discovered a number of genes that appear
to have been improperly annotated in biological databases (
Pseudomonas
Genome
Database, KEGG, NCBI). These mis-annotations arose due to a lack of information
at the time of the original annotation or because knowledge that was available in the
literature had been overlooked in the original annotation. In a number of other cases,
the model building process has also generated new hypotheses for gene functions. For
instance, our reconstruction process identified an unlikely gap in the L-lysine degra-
dation pathway of
P. putida
. Extensive literature search and careful reannotation has
provided considerable evidence that the genes PP0382 and PP5257, currently anno-
tated as “carbon-hydrogen hydrolase family protein” and “oxidoreductase, FAD bind-
ing” respectively, most probably code for a “5-aminopentamidase” and “L-pipecolate
oxidase”, respectively [36]. Another example is the propanoate degradation pathway:
In the iJP815
pre2
version this pathway was complete except for one enzymatic activ-
ity, namely the 2-methylisocitrate dehydratase. Analysis of the enzymes flanking this
reaction showed that all of the enzymes are encoded by genes immediately adjacent
to the ORF PP2330. Inspection of this region of the genome revealed that PP2336
is annotated as “aconitate hydratase, putative,” although the flanking genes are re-
sponsible for degradation of propanoate. Analysis of PP2330 via BLAST revealed a
homology of more than 99% over the whole length of the protein with the 2-methy-
lisocitrate dehydratase from other bacteria, such as other strains of
P. putida
(GB-1,
W619),
Burkholderia prymatum
STM 815,
Burkholderia multivorans
ATCC 17616,
Pseudomonas aeruginosa
PA7, and
Stenotrophomonas maltophilia
R551-3. Conse-
quently the gene was reannotated to code for this function and the gap in propanoate
degradation pathway was thus closed by addition of the corresponding GPR. In other
cases, discrepancies exist between various databases, as in the case of PP5029, which
is annotated in KEGG as “formiminoglutamase” but in NCBI as “N-formylglutamate
deformylase”. Analysis of network gaps, genomic context, and sequence homology
provided a strong indication that “N-formylglutamate deformylase” is the correct an-
notation. In many other cases the reannotation meant changing the substrate specific-
ity of the enzyme (which corresponds to changing the last part of the EC number).
These were mainly identified by BLASTing the protein against protein sequences of
other microbes and, whenever available, cross-checking the BLAST results against
primary research publications. The full list of reannotations suggested by the recon-
struction process is shown in Table 3.
