In PLOS Genetics, a Brown University-led team describes a neural network-based computational approach for mapping genetic associations and related genetic trait architecture from genome-wide association study annotation data. The "Biologically Annotated Neural Networks" (BANNs) method makes it possible to map SNPs and search for their over-representation, the researchers say — an approach they applied to simulated data and to data from a GWAS on several quantitatively tested traits in mice or humans assessed by the Wellcome Trust Centre for Human Genetics, the Framingham Heart Study, or the UK Biobank project. "By modifying a well-established variational inference algorithm, we are able to derive posterior inclusion probabilities which allows researchers to carry out SNP-level mapping and SNP set enrichment analyses, simultaneously," the authors write, noting that "the concept of partially connected neural networks may extend to any scientific application where annotations can help guide the groupings of variables."
A team from China and Pakistan presents findings from an analysis of candidate genes for major depressive disorder (MDD) proposed by prior genome-wide association studies. As they report in PLOS One, the researchers relied on a case-control approach involving some 400 individuals from northwestern Pakistan with MDD and 232 unaffected control individuals from the same population, comparing genotyping data for half a dozen candidate genes. Based on allele and genotype patterns at the marker SNPs, the authors did not see significant ties between MDD and the proposed genes. Even so, they say, while the study "does not support the major role of these polymorphisms in contributing to MDD susceptibility," they suggest that the work "does not preclude minor impact" and called for more complete analyses of these and other genes across larger cohorts.
Investigators at the Indian Council of Agricultural Research-National Research Centre on Equines compare genomic features within and between forms of Salmonella enterica serovars behind two distinct poultry diseases for another paper in PLOS One. Using several bioinformatic approaches, the team compared core and accessory gene features, phylogenetic relationships, and more for nine strains selected from the Gallinarum serovar and Pullorum biovar (bvP) of S. enterica, which has been linked to pullorum disease or from the fowl typhoid-causing biovar Gallinarum (bvG). "Comparative genome analysis unravels similarities and dissimilarities," they report, "thus facilitating identification of genomic features that aids in pathogenesis, niche adaptation, and in tracing of evolutionary history."