Elastin-like polypeptides (ELPs) are attractive materials for the green preparation of silica nanoparticles via biomimetic silicification. However, the critical factors affecting the ELP-mediated silicification remain unclear. Herein, the role of tunable amino acid residues of ELPs in silicification was studied using three ELPs (ELPs[V9F-40], ELPs[KV8F-40], and ELPs[K5V4F-40]) and their fusion proteins (ELPs[V9F-40]-SpyCatcher, ELPs[KV8F-40]-SpyCatcher, and ELPs[K5V4F-40]-SpyCatcher) with different contents of lysine residues. Bioinformatics methods were employed for the first time to reveal the key physicochemical parameters correlated with silicification. The specific activity of ELPs was increased with the promotion of lysine content with a high correlation coefficient (R = 0.899). Furthermore, exogenous acidic protein SpyCatcher would hinder the interactions between the silica precursors and ELPs, leading to the significantly decrease in specific activity. The isoelectric point (pI) of ELPs presented the highest correlation to silicification with a coefficient of 0.963. The charges of the ELPs [K5V4F-40] at different pH were calculated based on the sequence or structure. Interestingly, the excellent correlation between charges based on structure and specific activity was obtained. Collectively, the novel methods developed here may pave a new way for rational design of ELPs or other peptides for efficient and green preparation of silica nanomaterials for biomedicine, biocatalysis, and biosensor.