For this phase of the project, where we're casting a fairly wide net, results in the same signature class (shorter or longer signatures, subset of biomarkers, etc.) are gathered together to form a statistical picture of which markers and which combinations of biomarkers are successful at predicting or distinguishing cancer.

We're measuring the overall distribution of good signatures, but also looking at the top signatures and the top-scoring markers themselves. Our lab specializes in biological-network analysis, and so the top signatures and markers are then placed in their network contexts to get a bigger picture of why these genes or proteins are standing out. The same analysis will also tell us what other genes or proteins are close interactors (neighbours) in the network.

We use that information to optimize cancer biomarker signatures used in diagnosis or prognosis, and also to learn more about some of the mechanisms of cancer.

Christian A. Cumbaa
Research Associate
Ontario Cancer Institute / University Health Network

Hello there (if someone around here) !

I recently talked to a cancer researcher and he told me that enough cancer markers are known today and for him this project isn't useful today.

What is your opinion ? Have we enough markers ?

Hello Pablo,
thank you very much for your comment and giving us the chance to reply.

Over recent decades, numerous studies used diverse "omic" platforms to identify signatures for detection and prognosis of different cancers.
Differences in analytical methods and in characteristics of patient cohorts obviously led to a diverse set of markers.
Moreover, current statistical and machine learning algorithms only find a single signature per study, while for each study, one could find thousands of signatures with similar predictive power. From a biological and clinical point of view, it is beneficial to know multiple signatures - so we can improve our understanding of the disease, and also generate better/easier markers for clinical applications.

A cancer signature comprising a set of specific genes may appear different than another signature composed of another set of genes, and yet perform equivalently in terms of clinical value, because the genes may perform similar function.

In addition, translating the proposed signatures into clinical practice remains challenging; they often do not validate in other cohorts or by different biological assays, and there are thousands of possible combinations to consider.

At the present time, there is no appropriate approach to find all good sets of markers. Our knowledge about cancer markers is still too limited to diagnose all cancers, identify patient risks and predict response to treatment.
The understanding of underlying mechanisms could lead to better medication and most of all to more appropriate treatment choices for individual patients.

The first goal of the Mapping Cancer Markers (MCM) project is to gain a deeper understanding of the "rules" of why and how individual genes contribute to prognostic and predictive signatures, which is essential to understand the cancer biology. Second, identifying these multiple signatures, and performing integrative, network-based analysis - we aim to identify better (heuristic) algorithms to identify multiple signatures, which would be less computationally intensive.
Therefore, we are using the unique computational resource of the WCG to systematically survey the landscape of useful cancer gene signatures for multiple cancers (diagnosis and prognosis), and thereby establish a benchmark (gold standard sets) for cancer gene signature identification and validation.
Subsequently, we will utilize these patterns to identify generalized signature families that give deeper insights to the molecular background of cancers, and give rise to more reliable signatures for cancer detection and prognosis.
Furthermore, these results can verify novel heuristics to discover more generalized signature families in other areas of cancer research.

More details are highlighted in previous progress reports and the following scientific publications

Best wishes,
Your Mapping Cancer Markers Scientist Team


Boutros PC, Lau SK, Pintilie M, Liu N, Shepherd FA, Der SD, Tsao MS, Penn LZ, Jurisica I. Prognostic gene signatures for non-small-cell lung cancer. Proceedings of the National Academy of Sciences. 2009 Feb 24;106(8):2824-8.

Shedden K, Taylor JM, Enkemann SA, Tsao MS, Yeatman TJ, Gerald WL, Eschrich S, Jurisica I, Giordano TJ, Misek DE, Chang AC. Gene expression–based survival prediction in lung adenocarcinoma: a multi-site, blinded validation study. Nature medicine. 2008 Aug;14(8):822.

I don't pretend to understand all of the explanation, but thanks for it anyway and for the reference to the two articles.
Cheers

I expect you would have to convince the researchers that you could contribute to their efforts, and the code is probably not divulged lightly. But everyone would love a GPU version, though the science may not make that possible.
https://www.worldcommunitygrid.org/about_us/viewNewsArticle.do?articleId=571