Date: Fri, 29 Mar 2024 11:33:26 +0000 (UTC) Message-ID: <360754954.23.1711712006313@426d26ca652a> Subject: Exported From Confluence MIME-Version: 1.0 Content-Type: multipart/related; boundary="----=_Part_22_110812635.1711712006313" ------=_Part_22_110812635.1711712006313 Content-Type: text/html; charset=UTF-8 Content-Transfer-Encoding: quoted-printable Content-Location: file:///C:/exported.html
We are continuously evaluating potential postdoctoral researchers to joi= n our interdisciplinary data science team, spa= nning multiple research areas in ecology and evolutionary biology. The post= doctoral researchers will join a collaboration among eight faculty at the U= niversity of Wyoming, University of Montana, and University of Nevada-Reno,= including Drs. Alex Buerkle, Christopher Weiss-Lehman, Lauren Shoemaker, <= span class=3D"inline-comment-marker" data-ref=3D"803ca107-8fba-4bc1-97e0-76= 094b6fab16">Sarah Collins, and Daniel Laughlin (UW), Joanna Blaszcza= k and Matt Forister (UNR), and Bob Hall (UM). The team includes 13+ postdoc= toral researchers at a time.
The volume and availability of data have increased enormously over the l= ast decade in the life sciences. This expansion is rapidly changing the sta= ndard scale of analyses and is leveling and extending access to a broader p= opulation of scientists and the public via open-access data. Tremendous opp= ortunities for discovery are being newly created, but these opportunities b= ring with them substantial computational challenges. In the life sciences, = measuring many covariates for each unit of sampling is now common, providin= g highly dimensional measurements that could be used as variables in predic= ting biological patterns and processes.
A principal challenge arising from the data revolution is to maximally a= nd efficiently use information in available data and develop mechanistic ex= planations of the underlying biological processes (Fig. 1). The ideal is to= obtain a best fit of a model to the particular data, without including spu= rious relationships (i.e. =E2=80=9Coverfitting=E2=80=9D) that would not app= ly to other data, and to obtain highly interpretable models that are requir= ed for formalization and generalization in the advancement of science. Wher= eas the data revolution has spurred great advances in machine learning and = statistical modeling, this ideal remains elusive. We propose to address thi= s persistent gap in data science. We will develop innovative, hybrid analyt= ical methods that will allow the development of explanatory, mechanistic mo= dels while incorporating the high dimensionality of contemporary data. We w= ill confront our modeling approaches with existing experimental and observa= tional data from multiple biological domains, learn the support for differe= nt generative processes, and advance theory in the life sciences.
Figure 1: a) The scale of life science data has revolut= ionized our ability to assess the contribution of processes (covariates) to= biological patterns. b) More complex models trade-off between model fit to= current data (green) and predictive accuracy in new conditions (blue). c) = Interpretability decreases with the number of parameters. We will minimize = these tradeoffs and maximize interpretability (grey).
These fundamental gaps in analytical capacity are mirrored in scientific= training and the workforce. Scientific progress is impeded because knowled= ge of computational tools is insufficient for broad use due to gaps in trai= ning in computational and data science. Moreover, rapid methodological chan= ges in response to the data revolution have stymied communication across di= sciplines in the life sciences. We will address these needs, for scientists= at multiple career stages from undergraduate to faculty, build on the exis= ting activities of the Data Science Center at the University of Wyoming (UW= ), and extend these to the collaborating institutions and beyond. This will= include bolstering the success of applicants to the workforce, through: me= ntoring trainees for successful job searches and interviews, enhanced knowl= edge of attractive careers, and effective cross-disciplinary training in da= ta science.
The overall aim of the current proposal is to establish a consort= ium that will fundamentally advance our capacity to analyze challenging, hi= ghly dimensional data in the life sciences. This aim will be achieved throu= gh three complementary and integrated specific objectives:
Assess, develop, and disseminate innovative computational method= s for predictive and explanatory models of high-dimensional data. = Drawing on existing data science methods, we will compete models using real= and simulated data to assess their performance, extend and develop new met= hods using a hybrid modeling approach to increase both the interpretability= and predictive accuracy of models, and disseminate our results through wor= kshops, freely available published code, and publications.
Formalize life science knowledge in models for three cross-scale= domains. We will apply predictive and explanatory models to build= process-based understanding of the mapping between genomes to phenotypes, = the ecology of species-rich communities and global-scale species distributi= ons, and the interplay of temporal scales in aquatic ecosystem ecology. Our= consortium will share effective strategies and iteratively refine models a= nd computational solutions to efficiently advance knowledge across all thre= e domains.
Train and foster the development of scientists for the workforce= , through education and inclusion in data science. We will train 1= 2 postdoctoral researchers in cutting edge data science, combined with doma= in expertise. Our consortium will foster the development of novel, cross-di= sciplinary collaborations through annual meetings and regular communication= . We will provide regular training and professional development to advance = the careers of all participants.
While machine learning and computational methods have dramatically chang= ed our capacity for obtaining well-performing predictive models from high d= imensional data, many of these methods lack the diagnostic methods and tool= s to make them interpretable and allow the extraction of general, scientifi= c knowledge that can be applied across systems. This proposal will fund a c= onsortium of researchers who will work across life science disciplines to f= ind, assess, and further develop computational methods that will yield maxi= mally predictive and interpretable models. The consortium will facilitate t= he careers of several recently appointed professors and a large cohort of p= ostdoctoral researchers, in addition to enabling broad data science educati= on.
The University of Wyoming has strong research programs in ecology and ev= olutionary biology across multiple departments, including Botany, Zoology a= nd Physiology, Ecosystem Science and Management, Plant Sciences, and the Pr= ogram in Ecology. The university is located in Laramie, a community th= at is nestled between the Laramie and Snowy Mountain ranges, which offer am= ple opportunity for skiing, climbing, hiking, and mountain biking. Laramie = has a relatively low cost of living, is close to field sites across a wide = variety of vegetation types from mixed grass prairie to alpine tundra, rive= rs and lakes, and is within easy driving distance of Colorado=E2=80=99s Fro= nt Range corridor (Fort Collins, Boulder, and Denver).
The University of Nevada, Reno has strong research programs in ecology, = evolution, and conservation biology as well as hydrology across campus, inc= luding the Dept. of Natural Resources and Environmental Science, Biology De= pt., Dept. of Math & Statistics, and the Global Water Center. The unive= rsity is at the intersection of the eastern Sierra Nevada Mountains and the= Great Basin with incredible access to field sites as well as outdoor recre= ational activities including skiing, climbing, hiking, and mountain biking.= We are ~45 minutes from Lake Tahoe, three hours from Yosemite National Par= k, and 3.5 hours from San Francisco. The Reno community is also a diverse a= rtistic and cultural community with much to offer.
Flathead Lake Biological Station (FLBS) is year round field station and = part of the University of Montana, Missoula. FLBS is located on Flathead La= ke, a spectacular environment one hour south of Glacier National Park and 1= .5 hours north of Missoula. The Bio Station has full time faculty and staff= with research strengths in limnology, oceanography, and conservation genom= ics. Nearby sites range from glacial ecosystems to Flathead Lake and includ= e a long term research site on a connected river-floodplain on Middle Fork = Flathead River. Outdoor recreation opportunities are extensive and include = skiing, cycling, rafting, fishing, hiking, and backpacking in one of the la= rgest wilderness areas in the continental US.
Everyone in the consortium will have access to excellent shared computin= g resources in the Advanced Research Computing Center at the Unive= rsity of Wyoming.
The University of Wyoming, University of Montana, and University of Neva= da are Affirmative Action/Equal Opportunity Educator and Employers. We are = committed to a multicultural environment and strongly encourage application= s from women, minorities, veterans and persons with disabilities. Ple= ase see each institution=E2=80=99s individual advertisements for details.= p>