Research

Research directions

I work on immune-repertoire analysis and probabilistic drug-response modeling at ComputImm Laboratory, supervised by Prof. Changiz Eslahchi.

01 · Computational immunologyOngoing research

B-cell receptor lineage reconstruction

Question: What can receptor sequences tell us about the relationships and ancestry of B cells?
Approach: I investigate computational methods for reconstructing BCR lineage trees, with attention to how sequence relationships and modeling assumptions affect the interpretation of inferred ancestry.
Current status: Method development and research in progress. My master’s thesis topic is still being defined.

Related software: ClonoDefs

ClonoDefs makes repertoire grouping definitions explicit and produces traceable assignments. Its optional distance trees describe observed sequences; lineage reconstruction is a separate research question.

Implementation & grouping example

02 · Biomedical machine learningOngoing research

Drug-combination doses and inhibition

Question: How do two drugs, their doses, and cellular context relate to measured inhibition?
Approach: I use drug and cell-line information to study complementary forward and inverse models of this relationship.

Forward: doses → inhibition

Estimate response from a drug pair, paired doses, and cellular context.

Research in progress

Inverse: inhibition → doses

Model a distribution of possible dose pairs for a requested response.

Experimental implementation: DuoDose

Current status: DuoDose implements the inverse model and documents retrospective assessments. Forward modeling remains ongoing. Low reported target-hit rates motivate further work on generalization and response evaluation.

DuoDose architecture & evidence

This is methodological research using retrospective assay data. The models do not establish response at unmeasured doses or support clinical dose recommendations.