For a variant to have a causal effect on an organism-level trait, there must be a chain of causal events starting with the DNA-sequence, proceeding through one (or more often many) molecular intermediates in a functional pathway before it is observable at the organism level. The causal influence of a variant on a trait is, unfortunately, neither necessary nor sufficient for a variant to appear to be statistically associated with a trait in an association study, and spurious association of genotype and phenotype is not uncommon. It is for this reason that much of the hard work of a genetic association study begins after the association statistics have been generated. The true goal of the genetic association study is not simply to identify the genetic variation that most correlates with phenotype, but to try to identify the set of variants whose correlation with the trait of interest are driven by causal relationships, rather than by coincidence or confounding. In this dissertation I discuss three strategies for relating genotype to phenotype based on the results from genetic association studies. I first discuss the method FGEM, which combines the output from gene-basedassociation tests with gene-level annotation data to both estimate the enrichment of the annotations and re-prioritize genes based on those enrichment estimates. I find that FGEM's joint modeling of gene-level association data with gene-level annotation data is a powerful approach for identifying enriched pathways. Furthermore, I find that identification of enriched pathways can be used to identify additional causal genes. Next I describe my method for heritability estimation from summary statistics, RSSp. RSSp uses GWAS summary statistics and an estimate of pairwise Linkage Disequilibrium (LD) to estimate narrow-sense heritability. I find that RSSp estimates heritability in polygenic traits from GWAS summary statistics and a reference LD panel with accuracy comparable to in-sample methods. Finally, I discuss my efforts in discovering risk genes for preterm birth via fine-mapping GWAS summary statistics. I find that disease-relevant functional genomic annotations are useful for improving statistical fine-mapping. Using this approach I identified new genes not (directly) implicated from GWAS alone.