The Deep Core: Mapping the 0.91% Regulatory Backbone of the Human Proteome and Its Role in Cancer Drug Resistance

We present a computational method to identify the minimal regulatory backbone of the human proteome using multi-distance dominating set analysis. Applying our DeepDS algorithm to the human protein-protein interaction network, we identified a Deep Core of 144 proteins (0.91%) that achieve complete topological control of the network within 3 interaction steps. Comparative analysis across 7 species reveals that topological efficiency is an evolution- ary trait: humans (0.91%) are significantly more centralized than mice (3.29%), revealing a ”Modularity Gap” of 3.65×. Only 12 proteins are topologically conserved between mam- mals (”The 12 Immortals”), representing the irreducible machinery of life including EGFR, PLK1, and CASP3. We demonstrate that cancer preferentially hijacks this Deep Core. Intersection with clinical data reveals that 13.2% of the core consists of cancer drivers, including the ”Sinister Six” (TP53, EGFR, PLK1, SRC, SP1, CEP290). Finally, we validate a network-based mechanism of drug resistance. Statistical analysis of pharmacogenomic data (N = 321) reveals a strong correlation (r = 0.764, p < 0.0001) between high GNAL expression and Erlotinib resistance.