Detecting the existence of structure in a population is a preliminary step required for studies in several important areas of ecology and populations genetics, this includes studies aimed at
1) discovering genetic bases of diseases,
2) detecting DNA regions of functional importance,
3) retrieving historical demographic events,
4) identifying factors affecting migrations and gene flow
5) defining conservation genetic units of endangered populations or species.
As of today, these task involved mostly clustering algorithms, i.e. methods that attempt to group individuals into genetically homogeneous units. The three most used methods have been extremely successful to address questions 1-5 (receiving more than ten thousand citations in the last ten years, cf. Excoffier and Heckel 2006, Nature Rev. Genet., Guillot et al. Mol. Ecol., 2009). These methods are highly computer demanding though, requiring about an hour of computations for datasets consisting of a few tens of genetic markers. But while the increase of computers and algorithms speed has been of the order of a few per-cents in the last 3-4 years, the number of genetic markers available in typical population genetic studies has been multiplied by 1000-10000 in the same time. The community is therefore facing a problem of unprecedented difficulty to exploit current high-throughput datasets. This is a formidable statistical and computational challenge.
This PhD project is aimed at investigating how recent methods proposed in mainstream computational statistics can be modified and applied to population genetic data. We will specifically address the following points:
developing a new method to infer fine-scale spatial genetic structure,
speeding-up inference algorithms bringing down current unpractised computing times from several weeks or month to a few minutes or hours,
developing a computer program to disseminate the newly developed method into the scientific community,
testing the operational level of the method and applying it to a diversity of biological systems using marine fish species.
Carrying out these tasks is a necessary step to make full use of modern high-throughput data and to bring population genetics one step ahead. It will have applications in evolutionary biology in the broadest sense. We strongly believe that this aim is out of reach of pure biologists or pure statisticians, and we propose to build upon past on-going collaborations between scientists of the various fields in Denmark and France. The project above is suitable for a PhD or a postdoc and is fully funded under the Danish-Brazilian agreement Science without borders and the Informatics and Mathematical Modelling Department (IMM) at the Technical University of Denmark (DTU).
For PhDs, the package includes
a cutting-edge training program in statistics
a stipend (about 1800 Euro/month after tax for a PhD student)
a compulsory and fully funded stay in a research department outside Denmark
The project is also suitable for a postdoc ( information about salary and project duration available upon request). This funding opportunity is reserved to Brazilian citizens.
The scientific description above is only indicative. The balance between theory and application can be modified according to the interest and skills of the applicant.
Students with a MSc or PhD in statistics or related field (applied mathematics, bioinformatics, computer science) interested should send an email with a letter of motivation, a cv and the form
available from http://www.dtu.dk/upload/administrationen%20-%20101/afi/phd/grades_uk.xl...
carefully filled to
Department of Informatics and Mathematical Modelling
Technical University of Denmark, Copenhagen, Denmark
Office phone +45 45 25 33 21
Deadline : The final deadline for formal application is 31/01/2013. To comply with this deadline, potential applicants should contact G. Guillot in the very first week of january or earlier.