BACKGROUND: Colorectal carcinogenesis involves dynamic interactions between genetic susceptibility and cellular heterogeneity, yet current studies rarely disentangle causal genes from passive associations. While GWAS have mapped numerous risk loci, only a minority colocalize with eQTL/pQTL. A multi-omics framework combining single-cell transcriptomics, transcriptomics, proteomics, and MR is urgently needed to resolve cell-type-specific drivers of colorectal cancer pathogenesis. METHODS: We integrated GWAS data, eQTL data, pQTL data, and single-cell RNA sequencing differential gene expression profiles from public databases. Subsequent batch Two-sample Mendelian randomization and further SMR analysis aimed to identify key genes in the pathogenesis of colorectal cancer. RESULTS: Cluster analysis identified 4909 DEGs across various cell types. We discovered that 428 DEGs had a causal association with colorectal cancer through eQTL, of which 38 genes met the FDR statistical standards, and four of these genes (CTSF, PCSK7, LYZ, LMAN2L) also had causal associations through pQTL. SMR analysis confirmed the reliability of PCSK7 as a disease target. CONCLUSION: By integrating single-cell data, transcriptomic data, proteomic data and GWAS data for MR analysis, we identified CTSF, PCSK7, LYZ, LMAN2L as potential targets for colorectal cancer.