Epistasis Blog

From the Artificial Intelligence Innovation Lab at Cedars-Sinai Medical Center (www.epistasis.org)

Thursday, August 16, 2012

Journal Impact Factor

There has been a lot of discussion online the last few days about the value of journal impact factors for judging the quality of publications. This is important because faculty promotion and tenure is often measured by impact. The following are several posts calling into question the use of journal impact factors.

1) Nature editorial from 2005 titled "Not-so-deep impact: Research assessment rests too heavily on the inflated status of the impact factor"

2) A 2008 note about impact factors from the Editor-in-Chief of Nature title "Escape from the impact factor"

3) Another Nature editorial from 2010 titled "Dissecting our impact factor"

4) A PNAS editorial from 2010 that says "placing too much emphasis on publication in high impact factor journals is a recipe for disaster"

5) 2010 opinion piece in Front. Psychology titled "Are scientists nearsighted gamblers? The misleading nature of impact factors" - be sure and read the criticism of this piece in the comments by Pep Pàmies. The numbers presented seem suspect.

6) A provocative blog post by Stephen Curry titled "Sick of impact factors"

7) A blog post by Tom Webb titled "My own personal Impact Factor"

8) Mendeley page by Jonathan Eisen on papers that discuss impact factors.

9) The San Francisco Declaration on Research Assessment (DORA) released in 2013

10) Editorial in Science by Bruce Alberts from 2013 commenting on DORA

11) Editorial in The EMBO Journal from 2013 commenting on DORA

My personal view is that we should judge faculty based on three measures. These were originally suggested by Dr. John Blangero.

1) Total Citations. This is a good measure of impact. The more your papers are cited, the better your impact. This especially true for your first/senior authors papers.

2) Total Publications. This is a good measure of how hard you work. It takes time to write, submit, revise and publish papers.

3) Competitive Funding. This is a good measure of how your peers view your work. It is tough to get a grant funded from the NIH or NSF if you are not doing timely, innovative and significant work with solid methods.

Wednesday, August 15, 2012

Epistasis in Yeast

Leonid Kruglyak has put his latest paper on arXiv for comment. This is a well-written paper that explores additive and non-additive genetic effects for a variety of traits in yeast. The discussion of the results is comprehensive and informative. I would love to see this work extended to higher-order gene-gene interactions and gene-environment interactions. We need more work like this as we try to understand the complexity of genetic architecture. It is interesting to note that there has been some discussion about putting papers on arXiv for public viewing prior to publication. See this Nature News piece for an introduction to the topic.

Finding the sources of missing heritability in a yeast cross

Joshua S. Bloom, Ian M. Ehrenreich, Wesley Loo, Thúy-Lan Võ Lite, Leonid Kruglyak


For many traits, including susceptibility to common diseases in humans, causal loci uncovered by genetic mapping studies explain only a minority of the heritable contribution to trait variation. Multiple explanations for this "missing heritability" have been proposed. Here we use a large cross between two yeast strains to accurately estimate different sources of heritable variation for 46 quantitative traits and to detect underlying loci with high statistical power. We find that the detected loci explain nearly the entire additive contribution to heritable variation for the traits studied. We also show that the contribution to heritability of gene-gene interactions varies among traits, from near zero to 50%. Detected two-locus interactions explain only a minority of this contribution. These results substantially advance our understanding of the missing heritability problem and have important implications for future studies of complex and quantitative traits.

Monday, August 13, 2012

Modular Biological Complexity

There is a great new piece in Science on "Modular Biological Complexity". The author argues that without modules biology would be too complex to understand because it would take an eternity to run the experiments on all the many interactions. Note the idea of the "complexity brake".

Koch C. Systems biology. Modular biological complexity. Science. 2012 Aug
3;337(6094):531-2. [PubMed]

"Given the large number of components that cannot be averaged away, any possible technological advance is overwhelmed by the relentless growth of interactions among all components of the system. It is not feasible to understand evolved organisms by exhaustively cataloging all interactions in a comprehensive, bottom-up manner."

Monday, August 06, 2012

New NIH R01: Bioinformatics Approaches to Visual Disease Genetics

My new NIH R01 from the National Eye Institute (NEI) was funded starting August 1, 2012. The details can be found on the NIH RePORT website. The abstract is here:

It is now recognized that many visual diseases are influenced by complex interactions between multiple different genetic variants. As a result, our ability to predict susceptibility to visual diseases will depend critically on the computational, mathematical and statistical modeling methods and software that are available for making sense of high-dimensional genetic data. We propose here a bioinformatics research project to develop network modeling approaches for identifying combinations of genetic biomarkers associated with visual disease endpoints. Our working hypothesis is that a systems-based bioinformatics approach using network modeling will play a very important role in confronting the complexity of the relationship between genomic variation and visual diseases. We will first develop and evaluate modeling methods to infer large-scale genetic interaction networks from genome-wide association studies (AIM 1). We will then apply the modeling methods developed in AIM 1 to the inference of genetic interaction networks from genome-wide association data in subjects with and without visual diseases (AIM 2). Next, we will utilize the inferred genetic interaction networks to guide the development of predictive genetic models of visual diseases (AIM 3). Finally, all network modeling methods will be released to the vision research community as part of a popular user-friendly, freely available and open-source software package (AIM 4). We anticipate that the network modeling methods and software developed and distributed as part of this project will play an important role in the development of the genetic tests that will be necessary to identify those at risk for visual diseases.