DREAM Organizers Announce Cellular Network Prediction Challenge

YORKTOWN HEIGHTS, NY -The National Center for Biomedical Computing at Columbia University, in collaboration with the IBM (NYSE: IBM) Computational Biology Center at IBM T.J. Watson Labs, today announced a call for participation at its yearly workshop for the assessment of methods for reverse engineering of biological circuits.
The Dialogue for Reverse Engineering Assessments and Methods (DREAM) workshop is intended to provide a forum for researchers to discuss a framework for the evaluation of experimental and computational tools used to reverse engineer biological pathways.

DREAM2, the second-ever DREAM Conference, will feature the discussion of predictions of a set of five challenges and will be held at the New York Academy of Sciences on December 3-4, 2007.

DREAM organizers have posted five challenges inviting researchers to use computational methods to solve them. Participants to the challenges will be able to make predictions about the networks from which the data originated, blind to the actual networks. Organizers will disclose the networks and the researchers that produced the data during the DREAM2 conference.

The DREAM2 challenges consist of five datasets that were produced from biological or in-silico networks. Each challenge is composed of one or more dataset which originated from a network unknown to the participants to the challenge. The predictors are invited to infer as best as possible the network from which the data originated.

The challenges include:

* The BCL6-Targets Challenge. The challenge is to identify the actual targets of a transcription factor in mammalian B-cells, amongst a list of 200 genes containing both actual targets and decoys.
* The Yeast Protein-Protein Sub-network Challenge. The challenge is to infer the pairs of proteins that interact out of a list of 47 yeast proteins.
* The Five-Gene-Network Challenge. The challenge is to infer the network of interaction of five genes connected in a biological circuit that was transfected to an in-vivo organism, from unpublished gene expression and qPCR data.
* The In-Silico-Network Challenge. The challenge is to reverse engineer three in-silico generated networks, with qualitatively different topologies, from in-silico generated gene expression and proteomics data.
* The Genome-Scale Network Challenge. The challenge is to infer a piece of the gene regulatory network of an undisclosed model organism from high throughput gene expression data.

The DREAM2 workshop will address issues related to the identification of gold standards that could be used to test the accuracy of methods for reverse engineering of transcriptional, signaling metabolic and developmental networks. These gold standards could be either in-silico models of biochemical interactions or known biological pathways.

The workshop will foster dialogue on the advantage and disadvantage of the use of different types of high-throughput data for the task of reverse engineering and on the issue of biochemical validation of inferred interactions. Additionally, the workshop will focus on experimental designs, theoretical frameworks and computational algorithms for pathway inference, as well as on the definition of appropriate metrics for the evaluation of the quality of the predictions of reverse engineering methods.

Challenge participation and more information about the DREAM project can be found at: http://wiki.c2b2.columbia.edu/dream/index.php/The_DREAM_Project