The genotype-phenotype relationship, and its proper quantitative description, is a cutting edge challenge in biology. An important aspect of this relationship is understanding the impact of natural genetic variation on gene expression. One necessary step for this is the development of effective models of gene regulation that link DNA sequence information with gene expression level. We apply a sequence-level model of gap gene expression in Drosophila to analyze single nucleotide polymorphisms (SNPs) in 214 sequencedDrosophila melanogaster lines. We provide an analytical description of the influence of SNPs on gene expression in the framework of the chosen thermodynamic modeling approach. By quantifying this influence for each SNP via different measures, we demonstrate that regulatory effects from multipleSNPs combine almost linearly. The rate of the resulting SNP influence on expression correlates only weakly with the rate of its influence on the predicted binding affinity of the corresponding polymorphic binding sites, giving evidence that the ultimate effect that a SNP exerts on phenotype is effectively provided via the level of regulatory interactions between transcription factor binding sites (TFBSs). We find that TFBSs with strong effect on regulating the gap gene network exhibit reduced polymorphism. To detect the signal of natural selection, we contrast the patterns of polymorphism in the data to those in random computations for two phenotypic levels, the transcription factor binding affinity and the gap gene expression in the model. In agreement with some previous findings, we do not detect a significant contrast in these patterns for the binding affinities, but the difference for the gene expression level is highly significant. This difference is determined by lower variation in gene expression patterns in the lines and also expressed in smaller rate of the SNP regulatory influence accumulation with rising number of SNPs. In contrast, we find no statistically significant correlation between the SNP regulatory influence and its frequency in the population, which suggests that, due to the small effect that a single nucleotide mutation exerts on gene expression, selection acts only on their combinations. Taken together, these results provide a system-level view of how genetic variation translates to the level of gene expression.