Representative Projects From systems inference to validation-aware design

This project focuses on recurrent communication structures across cancer types, aiming to identify conserved regulatory axes and cancer-specific signaling niches with mechanistic interpretability.

• Built standardized scRNA analysis pipelines covering 10+ cancer types to characterize tumor ecosystems.
• Used Liana to identify conserved and cancer-specific ligand-receptor circuits.
• Reimplemented CoVarnet in Python and detected five modules related to cancer development.
• Integrated spatial transcriptomics data to localize signaling niches and validate sender-receiver spatial co-localization patterns.
• Defined cell hub populations based on topological metrics and linked hub-related pathways to phenotypic programs.

This project determines whether TP53RK function in medulloblastoma mainly depends on phosphorylation activity or KEOPS complex-related mechanisms through controlled perturbation and rescue experiments.

• Constructed TP53RK phosphorylation-site mutants and KEOPS critical-site mutants.
• Introduced mutant constructs through lentiviral transduction and degraded endogenous TP53RK using dTAG-V1.
• Used Incucyte live-cell imaging to compare proliferation and cell-death rescue effects across mutant types.

TasAnchor is a modular adhesion and immobilization platform built in Bacillus subtilis, integrating genome editing, scaffold design, and material surface functionalization.

• Knocked out endogenous TasA and constructed a pHT01 vector using a SpyTag/SpyCatcher scaffold.
• Cloned expression constructs into E. coli BL21(DE3), induced protein expression with IPTG, and purified target proteins by Ni-NTA affinity chromatography.
• Integrated adhesion modules on polystyrene substrates to support biofilm immobilization on reactor membranes.