Biological process linkage networks

Abstract

Background: The traditional approach to studying complex biological networks is based on the identification of interactions between internal components of signaling or metabolic pathways. By comparison, little is known about interactions between higher order biological systems, such as biological pathways and processes. We propose a methodology for gleaning patterns of interactions between biological processes by analyzing protein-protein interactions, transcriptional co-expression and genetic interactions. At the heart of the methodology are the concept of Linked Processes and the resultant network of biological processes, the Process Linkage Network (PLN).

Results: We construct, catalogue, and analyze different types of PLNs derived from different data sources and different species. When applied to the Gene Ontology, many of the resulting links connect processes that are distant from each other in the hierarchy, even though the connection makes eminent sense biologically. Some others, however, carry an element of surprise and may reflect mechanisms that are unique to the organism under investigation. In this aspect our method complements the link structure between processes inherent in the Gene Ontology, which by its very nature is species-independent. As a practical application of the linkage of processes we demonstrate that it can be effectively used in protein function prediction, having the power to increase both the coverage and the accuracy of predictions, when carefully integrated into prediction methods.

Conclusions: Our approach constitutes a promising new direction towards understanding the higher levels of organization of the cell as a system which should help current efforts to re-engineer ontologies and improve our ability to predict which proteins are involved in specific biological processes.

Figure 1. Proteins that participate in “main pathways of carbohydrate metabolism” (rectangular nodes) and interact with proteins that participate in “response to stress” (elliptic nodes).

The colors of the nodes represent the sub-term annotation: rectangular nodes correspond to: “tricarboxylic acid cycle intermediate metabolism” (yellow), “gluconeogenesis” (orange) and “glycolysis” (dark brown). Elliptic nodes correspond to “response to-”: “DNA damage stimulus” (bright blue), “osmotic stress” (dark purple), “heat” (dark green), “starvation” (bright green), “oxidative stress” (dark blue). 3 genes are not annotated to any process more specific than “response to stress' (pink). A total of 62 proteins are annotated with “main pathways of carbohydrate metabolism”, and interact with 190 proteins that are not annotated with “main pathways of carbohydrate metabolism”. Of the latter, 38 are annotated with “response to stress”. The total number of proteins annotated with “response to stress” is 337. Consequently, the PPI-link between the two processes is predicted (p-value: 1.47e⁻⁵).

Figure 2. A histogram describing the percentage of process links associated with different semantic similarities.

All three networks, show similar distribution, although the number of links is different in the three networks (21,097 links in the PPI-PLN, 48,844 links in the GI-PLN and 4,521 links in the expression-PLN).

Figure 3. A comparison between the average precision, average recall and average F-measurement obtained by the MV and the IMV methods (for D. melanogaster).

For each term the activation threshold was chosen to yield the best F-measurement. The average was taken over those GO-terms for which at least one True Positive was found. There are 551 such GO-terms when using the MV method, and 672 such GO-terms using the IMV method.

Figure 4. Prediction of “larval development” annotation for the gene CG10701.

The sub-network of the D. melanogaster PPI-network containing the genes CG10701, in the middle, and its direct neighbors. Neighbors that currently lack process annotations are black. Neighbors that are annotated with the GO-process “larval development” are blue. Neighbors that are annotated only with processes which were not found to have a PPI-link to “larval development” are red. The other neighbors are annotated with at least one process that was found to have a PPI-link to the predicted process. Purple: cell organization and biogenesis; Green: cell communication; Brown: locomotory behavior; Orange: morphogenesis of an epithelium.

Figure 5. The probability for two genes with known specific biological process (S. cerevisiae) to genetically interact.

(i) without prior knowledge; given that the genes participate in (ii) PPI-linked processes, (iii) expression-linked processes and (iv) processes that are both PPI-linked and expression-linked. The total number of genetic interaction that was considered here is 15,228.

Figure 6. Log-log plot of the number of pairs of linked processes, as a function of the p-value of the link (PPI-Links, S. cerevisiae).

Both types of random annotations yield far fewer pairs with very low p-values than the actual network.