To identify additional causal gene variants contributing to variance in obesity–associated traits, an additional advanced intercross line was generate using a different substrain of the BFMI (BFMI861-S1) which is also carrier of the obesity locus on chromosome 3 and in addition possesses the highest liver triglycerides compared to the other BFMI lines.
The identification or the assessment of the importance of candidate genes and their variants that contribute to the differences in body weight, liver weight and fatty liver will allow us in the future to identify individuals at risk of specific pathological phenotypes such as obesity, insulin resistance, type 2 diabetes and fatty liver.
General linear models were fitted to the trait values including subfamily, litter size and sex as fixed effects. Different statistical models were used to do the mapping of all phenotypes (y) onto the different genotype models: Body weight: y = sex + mother + marker genotype; Liver weight: y = sex + mother + marker genotype; Liver triglycerides: y = marker genotype. P-values were corrected for multiple testing using a Bonferroni correction. Correction for multiple testing was performed using the number of informative SNPs as the total number of tests performed. A logarithm (base 10) of odds (LOD) score after Bonferroni correction above 4.5 was deemed to be ‘genome-wide highly significant’ and above 4 was deemed ‘genome-wide significant’ when supported by multiple adjacent SNP markers. QTL regions were defined by a conservative 1.5 LOD drop from the top marker; region start and end positions are defined by the first markers upstream and downstream of the top position that have a LOD score 1.5 LOD lower than the top marker. Looking for additional effects outside the major effect QTL (jObes1) located on chromosome 3, we used a multiple QTL mapping approach.11 We adjust our single QTL model to compensate for the effect of the jObes1 locus by including the top marker from the chromosome 3 region (SNP UNC5048297) as an additional cofactor into the model: Y = sex + mother + UNC5048297 + marker genotype + error.
For prioritization, first all coding genes in the identified regions were downloaded using bioMART (R package). As a next step, all SNPs within these genes were listed using medium coverage sequencing data of one substrains of the BFMI which were compared to the reference genome B6N using BCFtools. The impact of these SNPs was defined using Ensembl Variant Effect Predictor and candidate genes were ranked according to their impact on protein structure (from highly deleterious to tolerated).