NEBULON- a system for the inference of functional relationships of gene products from the rearrangement of predicted operons

Sarath Chandra Janga¹, Julio Collado-Vides¹ & Gabriel Moreno-Hagelsieb²

¹Program of Computational Genomics, CIFN-UNAM, Apdo Postal 565-A, Cuernavaca, Morelos, 62100 Mexico. ² Department of Biology, Wilfrid Laurier University , 75 University Avenue West, Waterloo , ON , Canada , N2L 3C5.

Abstract

Here we introduce Nebulon, a system to build networks of predicted functional relationships of gene products based on their organization into operons in any available genome. The system is based on a previously developed method to predict operons by the distances between adjacent genes in the same strand, and on the high recombination rate of operon associations across genomes that reveal functional relationships among gene products. Our system can use different kinds of thresholds to accept a functional relationship, from those related to the prediction of operons, to finding the association in at least a given number of non-redundant genomes. We also work by shells, meaning that we decide on the number of linking iterations to allow for the complementation of related gene sets. The method shows high reliability benchmarked against knowledge databases of functional interactions. We also exemplify with several known and characterized functional gene sets.

Accessing Nebulon

Nebulon can be accessed by one of the two ways

a) The GUI can be accessed by following the link below.The GUI gives information for each link in nebulon, like source of the evidence,number of evidences and the functions of the genes in a given nebulome etc.For further information and help to use the interface please read the documentation of Nebulon-GUI.

http://tikal.cifn.unam.mx/nebulon

b) The command line toolkit of Nebulon provides a means to get the links between a given gene in a given genome under different thresholds.It is downloadable from here. We recommend using the command-line version if your interest lies in obtaining/analyzing interactions for large number of genes because for large outputs the GUI could be slow. Please read the documentation that comes with this tar-ball for any help in using the parameters and the program itself. The output of the interactions is a simple tab separated file and so could be used to view in cytoscape or yed.

Sections
1) Description of pathway identifiers used in Figure 4.

2) Distribution of Internal Vs External links for each Non-redundant genome in nebulon and the three letter codes for complete genomes used in command-line tool.

3) Table1 showing the newly found links in the recovery of argR regulon.

4) Interactions file in gml format for Figure 5.

5) Interaction files in gml format for Figures 6(a) and 6(b).

6) Interaction file in gml format for Figure 7 and Table 2 showing the genes having at least two links with genes related to Nitrogen Fixation in Sinorhizobium meliloti.

7) Table showing the number of functional links which are unique to nebulon(Found only by nebulon) and those which are also identified by STRING in different genomes (Only available for those genomes which can be mapped to entrez genomes database from string).

2. Graph showing the distribution of internal vs external links in all the NR genomes can be downloaded from here in Postscript format or in PDF format(Note: The figure has the genomes sorted into Proteobacteria,Firmicutes and Archea).Description of the three letter codes used can be found here and the Table which contains the link-proportions in each genome for internal and external links for the raw interaction data and at 0.4 log-liklihood threshold can be found here.The columns in this table represent the following

column 1 : Genome

column 2 : Number of internal links identified

column 3 : Number of Exernal links identified

column 4 : Number of links due to fusions

column 5 : Total number of links identified.(Note: A link can be identified by more than one means i.e internal,external or fusion so this need not be the commulative of the above)

column 6 : Fraction of links which are internal.

column 7 : Fraction of links which are external.

* Cases where we expect the genes to be linked functionally because the log-likelihood scores are high and the orthologs are conserved with no intervening genes in the genome of evidence. The genes gmk , ychE and yfjB have been predicted to be regulated by ArgR (Robison et.al , 1998) . It can also be noticed that in all these cases the genes are either putative, hypothetical or poorly annotated indicating the possibility of these associations to be real. In all 13 of these links we only expect the links marked ‡ (3 in number) to be false positives because of the high number of intervening genes. Such links could serve as a guide for future refinements in Nebulon. Complete genome names can be found in the web page.

4. Interactions in gml format for Figure 5 can be downloaded from here .This file can be viewed in cytoscape or yed.The same interactions with detailed information can also be obtained from the GUI of nebulon with the query gene as argR and query Genome as Escherichia coli K12 .

5. Interactions in gml format for Figure 6 and 7 can be downloaded from here.These files can be viewed in cytoscape or yed . The interactions with detailed information for figure 6(a) can also be obtained from the GUI or command line toolkit of nebulon with the query gene as tufA and query Genome as Escherichia coli K12 and those for Figure 6(b) can be obtained from with query gene as flgA , query Genome as Escherichia coli K12 and number of generations as 2 (Please note that currently Graphical interface doesn't support the number of generations threshold, so we recommend you to use command-line tool if you are interested in more than one generation queries.).

6. Interactions in gml format for Figure 7 can be downloaded from here.These files can be viewed in cytoscape or yed .

Table 2. Genes having at least two links with genes related to nitrogen fixation in Sinorhizobium meliloti.

7. Click here to see the table containing the details of the functional links identified by nebulon per genome. (Those which are unique to nebulon and those which are also found by string).

Note that in each case nebulon recovers significant number of links which are not found by string, irrespective of the fact that string uses a number of genomic context tools,High throughput experimental data, Coexpression data (which is normally from microarrays) and text-mining.

For Questions/Comments, please mail: sarath AT cifn.unam.mx