In the past 15 years, genome-wide association studies (GWAS) have identified thousands of genetic variants associated with a plethora of human complex traits, and greatly expanded our knowledge on their genetic architectures. Around 90% of GWAS variants are located in noncoding regions, suggesting that they may have regulatory effects. However, statistical integration of GWAS and expression quantitative trait loci (eQTL) can only explain ~25% of GWAS variants, highlighting the challenges in uncovering the molecular mechanisms underlying GWAS signals. Here, we show that genetic variants affecting pre-mRNA splicing (sQTL) can pinpoint GWAS loci missed in eQTL. Together, eQTL and sQTL explain 40% of GWAS variants. We demonstrate that primary, steady-state eQTL and sQTL are largely shared among immune cell types in peripheral blood, suggesting that it is challenging to pinpoint the exact cell type in which these shared QTL mediate disease risk. I also show that disease samples from the correct tissue can capture regulatory effects distinct from healthy or in vitro stimulated immune cells. We then investigate the regulatory effects on chromatin accessibility and relevance to disease. We find that chromatin accessibility (caQTL) greatly complement eQTL/sQTL in explaining disease GWAS. Through extensive comparisons with caQTL, eQTL and GWAS, I conclude that partially due to enhancer pleiotropy, even though caQTL can correctly reveal the effect of disease variants on chromatin accessibility, it is much harder to pinpoint the true causal gene and causal context when a caQTL has no effect on gene expression. This highlights the crucial importance to study more disease-relevant contexts and integrate caQTL with multi-modal annotation data for deeper understanding of the causal mechanisms of disease GWAS. Taken together, our results offer a more comprehensive understanding of the regulatory landscape of molecular phenotypes, especially for chromatin accessibility and gene expression, and their roles in the genetic predisposition of immune-related diseases. We highlight the complexity in layers of gene regulation cascade and provide a more cautionary tale in the interpretation of molQTL-GWAS data. We believe our work form a basis for future studies to further extend the boundaries of our knowledge of how genetic variants can shape human phenotypes through regulation of a network of molecular features.