Hypothesis testing related to the discovery of family relationship is still based on ancient models in need of urgent revision. This project aims to develop a new hypothesis testing approach based on mathematical parameters and the concept of identity-by-descent. Studies deriving the theoretical properties of the test are expected to be carried out. Once a hypothesis testing approach has been developed and properly evaluated, its scope can be spread to include missing data cases (due to DNA degradation leading to dropout) or distant family relations. The scientific community and statistics applied to forensic genetics are expected to benefit from the project. The donor partner, Norwegian University of Life Sciences, has many experts in the development of statistical methods applied to forensic genetics. The researcher from the project promoter did his Ph.D. in statistical genetics and has a bunch of publications in the field. The project will address the challenges with a multidisciplinary approach through the collaboration with many researchers of different knowledge areas.
Summary of project results
Statistical analysis of DNA data is nowadays a leading field. A great variety of applications can be found in the different branches of genetics including clinical, forensic or population genetics. The project aimed at developing a hypothesis test to detect family relationships (forensic genetics) with applications to other areas, such as civil law, clinical genetics or biomedicine. Presence of family relationships is on many occasions needed to know, and the advances in genotyping technologies and bioinformatics provide us with enough data to detect up to second-degree or, in some cases, third-degree relationships. Calculation of likelihood ratios for paternity and general relationship cases have largely been based on a procedure that lacks of mathematical parameters and without a proper study of its mathematical properties. As a consequence, we started a thorough revision of the bibliography which addressed us to the so-called identity-bydescent parameters, which describe family relationships in a specific way. Taking previous studies from Jacquard (1972) or Thompson (1975) as a basis, we developed a new hypothesis testing procedure based on mathematical parameters. Preliminary simulation studies were carried out in order to observe if the test was powerful enough, with successful results. Study of the mathematical properties of the test, essential for its consistency, were carried out together with Ignacio López-de-Ullibarri and Nils L. Hjort. A real dataset with 27 children coming from a case-control study of meningococcal disease was obtained from Antonio Salas, also co-author of the work; a thorough study was carried out looking for the presence of any relationship in the dataset. Comparisons with the classical method were established and more advanced simulations were carried out. This new testing procedure proves to be highly competitive regarding detection of family relationships, as indicated by testing simulations focusing on power. Once the “field” work is finished, the NILS applicant, starts to write a scientific article, which has been already submitted and it is now under revision. These have been the main steps (summarized) of the main research project, although parallel to this, different subjects have been dealt with. We think the main beneficiary of the project is the field of forensic genetics and, by extension, all knowledge areas which need, in some way, detection of family relationships: from civil law to clinical genetics.
Summary of bilateral results
Besides the joint research performed, partners have carried out additional joint activities. Several scientific papers are in preparation. Manuel García Magariños and the host institution researcher, Thore Egeland, have participated in September 2014 in the workshop conference “Statistical and computational methods for relatedness and relationship inference from genetic marker data” as invited speakers, in Edinburg, UK. Many researchers working in relatedness problems and pedigree reconstruction with different species were congregated there and our work had a work reception. Furthermore, the conference was important in order to meet other researchers in the field, such as Klaas J. Slooten. A collaboration with the Norwegian researcher Magnus Vigeland has started from this conference. Study of inbreeding relationships is a somewhat more complicated problem than the study of non-inbreeding relationships, as carried out in the two previous studies. This more difficult nature lies on mathematical and optimization grounds, as more parameters need to be estimated (to say so, with inbreeding the number of possibilities grows). Hence, the problem becomes measureless. The promoter has started to work with Magnus Vigeland, fellow mathematician at the Department of Medical Genetics in the University of Oslo. The work in question consists of developing a mathematical mapping which allows reducing the dimensionality when trying to detect inbreeding relations. Although it is still in a preliminary phase, we think the work is promising. Together with Magnus Vigeland and Thore Egeland are the three coauthors of this work. Provisional title is “Maximum likelihood estimation of IBD coefficients: violating the assumption of no inbreeding”. Once these works have been finished, it is the main aim of both the Spanish researcher and the host Norwegian researcher to work in the development of a software package implementing the hypothesis testing procedure. This will be done in cooperation with bioinformaticians and geneticists of the group of Antonio Salas. The purpose is to increase the popularity of the method especially in the forensic genetics field.