Interviews
Clinical and Computational Perspectives
As part of our Integrated Human Practices, we interviewed four medical doctors specializing in oncology and one professor specializing in in silico protein engineering. Their insights helped us understand the clinical needs, limitations of current therapies, and the opportunities and challenges of computational protein engineering.
Interview with Clinical Doctors
1. Based on your clinical experience, what do you think is the biggest challenge in treating cancer patients today?
Professor Yongsang Hong, Department of Oncology, Asan Medical Center:
Institutional barriers (such as reimbursement and regulatory restrictions), cultural barriers (distrust of doctors, rejection of clinical trials, reliance on unproven therapies), and medical barriers (advanced cancer is still incurable despite new drugs). Survival improvements are still limited to a few cancer types.
Professor Inkeun Park, Department of Oncology, Asan Medical Center:
Even with targeted and immune therapies, drug response varies widely due to genetic mutations and tumor microenvironment. Identifying predictive biomarkers is a key challenge. Economic barriers like drug costs remain significant.
Professor Choongryul Oh, Department of Hematology–Oncology, Chung-Ang University Hospitalhoongryul Oh, Department of Hematology–Oncology, Chung-Ang University Hospital:
Treatment complexity has increased as therapies multiply, but this also reflects the lack of a clear standard. While survival is improving, patients face prolonged side effects, reduced quality of life, and financial strain.
Professor Daeyoung Kim, Department of Oncology, CHA Bundang Medical Centeraeyoung Kim, Department of Oncology, CHA Bundang Medical Center:
Cancers require different immune targets, and tumors develop resistance mutations over time, requiring constant innovation.
2. How do you decide whether a patient is suitable for immunotherapy versus chemotherapy or targeted therapy?
Professor Yongsang Hong, Department of Oncology, Asan Medical Center:
We follow clinical guidelines based on large trials. For example, lung cancers with EGFR mutations receive targeted therapy; those with high PD-L1 expression may receive immunotherapy. Contraindications (e.g., autoimmune disease, organ transplant) are also considered.
Professor Inkeun Park, Department of Oncology, Asan Medical Center:
Molecular profiling is essential. We test for EGFR, ALK, ROS1 mutations in non-small cell lung cancer. If absent, PD-L1 expression guides immunotherapy decisions. NGS enables broader profiling for precision medicine.
Professor Choongryul Oh, Department of Hematology–Oncology, Chung-Ang University Hospitalhoongryul Oh, Department of Hematology–Oncology, Chung-Ang University Hospital:
Decisions depend on cancer type, stage, and molecular features. Clinical trial data inform whether immunotherapy is more effective alone, in combination, or not at all. Cost is another barrier.
Professor Daeyoung Kim, Department of Oncology, CHA Bundang Medical Centeraeyoung Kim, Department of Oncology, CHA Bundang Medical Center:
We consider prior research evidence, expected responsiveness, insurance coverage, and patient condition.
3. What do you see as the greatest limitations of current immunotherapies (checkpoint inhibitors, CAR-T)?
Professor Yongsang Hong, Department of Oncology, Asan Medical Center:
Only some patients respond, responses are hard to predict, and side effects can be severe and permanent.
Professor Inkeun Park, Department of Oncology, Asan Medical Center:
Checkpoint inhibitors are less toxic but effective only in a minority of patients. CAR-T is powerful but costly, time-consuming, and logistically complex.
Professor Choongryul Oh, Department of Hematology–Oncology, Chung-Ang University Hospital:
Mechanisms of checkpoint inhibitors are not fully understood. They often require combinations with other drugs. Side effects can be life-threatening. CAR-T takes too long to prepare for advanced patients, has serious side effects, and remains very expensive.
Professor Daeyoung Kim, Department of Oncology, CHA Bundang Medical Center:
Checkpoint inhibitors vary widely in response depending on cancer type. CAR-T requires suitable target markers, limiting applicability.
4. Our project focuses on engineering MHC Class I molecules to improve tumor antigen presentation. How closely is antigen presentation linked to the success or failure of immunotherapy?
Professor Yongsang Hong, Department of Oncology, Asan Medical Center:
Improved presentation can help, but tumors also evade immunity through other mechanisms. MHC engineering alone may not be sufficient.
Professor Inkeun Park, Department of Oncology, Asan Medical Center:
Antigen presentation is central. Enhancing MHC expression could improve outcomes, provided side effects are managed and biomarkers are developed to predict response.
Professor Choongryul Oh, Department of Hematology–Oncology, Chung-Ang University Hospital:
Antigen presentation is essential but not sufficient. Some patients respond strongly when many abnormal antigens are present, while others do not respond at all. Resistance mechanisms must also be addressed.
Professor Daeyoung Kim, Department of Oncology, CHA Bundang Medical Center:
Antigen presentation is critical, but modifying MHC Class I directly is conceptually challenging since it is also a personal identity marker.
5. If engineered MHC proteins were proven safe and effective, how might they improve current cancer treatment?
Professor Yongsang Hong, Department of Oncology, Asan Medical Center:
They could enhance outcomes, though exact roles would depend on trial results.
Professor Inkeun Park, Department of Oncology, Asan Medical Center:
They could stabilize antigen presentation and complement checkpoint inhibitors, boosting immune response.
Professor Choongryul Oh, Department of Hematology–Oncology, Chung-Ang University Hospital:
Most likely combined with PD-1 inhibitors for synergistic effect.
Professor Daeyoung Kim, Department of Oncology, CHA Bundang Medical Center:
Conceptually promising, but difficult to visualize without clear evidence.
6. Could this technology be realistically integrated into current protocols, or would it require a completely new approach?
Professor Yongsang Hong, Department of Oncology, Asan Medical Center:
Integration depends on large phase III trials. It may be added to existing protocols or stand alone, depending on delivery.
Professor Inkeun Park, Department of Oncology, Asan Medical Center:
If safe and effective, it could complement checkpoint inhibitors, but its role depends on comparative outcomes.
Professor Choongryul Oh, Department of Hematology–Oncology, Chung-Ang University Hospital:
It would require successful clinical trials to enter protocols.
Professor Daeyoung Kim, Department of Oncology, CHA Bundang Medical Center:
Hard to judge without a clear picture of how MHC Class I is engineered.
7. From the patient's perspective, what is the most important factor in adopting a new cancer therapy?
Professor Yongsang Hong, Department of Oncology, Asan Medical Center:
Patients want highly effective, safe, and easy-to-administer drugs, though expectations are often unrealistic.
Professor Inkeun Park, Department of Oncology, Asan Medical Center:
Safety first, then superior efficacy.
Professor Choongryul Oh, Department of Hematology–Oncology, Chung-Ang University Hospital:
Patients expect cures with fewer side effects, but all drugs involve trade-offs. Many therapies extend survival rather than achieve cures.
Professor Daeyoung Kim, Department of Oncology, CHA Bundang Medical Center:
Safety, then efficacy, then accessibility.
8. What potential side effects or unintended outcomes should researchers focus on?
Professor Yongsang Hong, Department of Oncology, Asan Medical Center:
Side effects must be well understood before trials. Preclinical and phase I testing are crucial.
Professor Inkeun Park, Department of Oncology, Asan Medical Center:
Immune-related toxicities should be prioritized.
Professor Choongryul Oh, Department of Hematology–Oncology, Chung-Ang University Hospital:
Engineered proteins may affect healthy cells. Preventive strategies are needed.
Professor Daeyoung Kim, Department of Oncology, CHA Bundang Medical Center:
Most drug candidates fail due to safety or efficacy issues, with enormous costs.
9. If you were advising a student team, what should we emphasize in terms of clinical relevance?
Professor Yongsang Hong, Department of Oncology, Asan Medical Center:
Focus on how you might overcome current immunotherapy limitations, not on clinical trial planning.
Professor Inkeun Park, Department of Oncology, Asan Medical Center:
Explain immune evasion and how your approach addresses it.
Professor Choongryul Oh, Department of Hematology–Oncology, Chung-Ang University Hospital:
Emphasize how your work might overcome resistance and combine with existing drugs.
Professor Daeyoung Kim, Department of Oncology, CHA Bundang Medical Center:
Frame your project as a proof-of-concept, recognizing the long path to clinical application.
Interview with Protein Engineering Expert
10. From your perspective, what are the benefits and limitations of relying on in silico predictions (like docking or AI models) when deciding mutations for MHC Class I?
In silico predictions are powerful for screening mutations and peptides quickly, avoiding random trial-and-error experiments. They show how mutations might affect stability or structure, and AI models can capture subtle sequence patterns humans miss. They save time and resources and help integrate structure, sequence, and binding data.
However, limitations are clear. Docking treats proteins as static, while MHC Class I is dynamic, with grooves opening and closing depending on peptide. Scoring is simplistic and often ignores water molecules, hydrogen bonds, and free energy changes. AI models depend on existing data and may be biased toward common alleles and peptides. They rarely capture interactions among multiple mutations. In the end, predictions are best as candidate generators, not final answers.
11. How reliable are in silico predictions for guiding MHC Class I mutations?
They are highly reliable as a starting point, narrowing thousands of mutations to a manageable list. AI models provide valuable structural and sequence insights, highlighting patterns invisible to human intuition. While they cannot replicate all biological processes, modern algorithms are increasingly accurate, and when paired with experiments, they guide efficient and systematic discovery.
12. How do you balance computational prediction with experimental validation?
The best approach is iterative. AI predictions reduce experimental load by filtering candidates, but experimental testing is essential. Usually, 10--20% of top candidates are validated experimentally to assess prediction accuracy. Results then refine the models, creating a cycle: prediction → validation → improvement.
13. What additional computational steps would strengthen predictions?
Molecular dynamics (MD) simulations add realism by capturing how MHC--peptide complexes move and change over time. RMSD, RMSF, and hydrogen bond persistence can be tracked to measure stability. MD also reveals allosteric effects and groove flexibility not seen in docking. Cross-validating multiple AI models and incorporating network analysis can further improve reliability.
14. Does our workflow (AI prediction → E. coli expression → binding assays) seem like a solid pipeline for high school-level synthetic biology?
Yes. It is simple, efficient, and meaningful. AI predictions narrow candidates, bacterial expression allows protein production, and binding assays provide validation. For stronger results, cross-check predictions with MD simulations or conservation analysis. Alternative systems like cell-free expression could also be considered, though not essential. Overall, this is an excellent learning pipeline for a student team.
Insights from Expert Interviews and Impact on Our Project
Speaking with both clinical specialists and a computational protein engineering expert gave us valuable feedback that has helped refine our project goals and experimental plans.
1. Clinical Perspectives
From the oncologists' responses, it became clear that one of the biggest challenges in cancer treatment is the variability in patient responses. Even with powerful immunotherapies like checkpoint inhibitors or CAR-T cells, only a subset of patients benefit, and side effects can be severe. They emphasized that antigen presentation is indeed a critical step, but also warned that simply enhancing MHC presentation will not fully overcome cancer's ability to escape immune surveillance. Tumor heterogeneity, resistance mechanisms, and patient-specific biology remain major barriers.
Their advice has led us to frame our project not as a "complete solution" to cancer immunotherapy, but as a potential enhancer that could make existing therapies more effective. We will highlight how our engineered MHC could complement checkpoint inhibitors or cancer vaccines rather than replace them. This feedback also reminded us to consider long-term issues such as safety, immune-related side effects, and the need for biomarkers to identify which patients would benefit most.
2. Computational Perspectives
From the in silico protein engineering expert, we learned that computational predictions are powerful but limited tools. Docking and AI models can guide us to promising mutations and save experimental resources, but they cannot capture all the dynamic and cellular complexities of MHC Class I. He stressed the importance of combining predictions with experiments in a "predict → test → refine" cycle. Additional computational steps like molecular dynamics simulations could strengthen predictions, and validating at least a portion of predicted candidates experimentally is essential.
This advice confirmed that our pipeline (AI prediction → E. coli expression → binding assays) is a strong starting point for a high school iGEM team, but also encouraged us to incorporate cross-validation and feedback loops to make our approach more rigorous.
3. Integration into Our Project
Based on these insights, we are adjusting our project narrative and design in the following ways:
- We will emphasize antigen presentation as one bottleneck, but clearly acknowledge other immune escape mechanisms and present our work as part of a broader immunotherapy toolbox.
- We will position our engineered MHC not as a universal cure, but as a strategic enhancer of current therapies like checkpoint inhibitors, cancer vaccines, or adoptive T-cell transfer.
- We will explicitly include a DBTL cycle that integrates computational predictions with experimental testing, showing that we understand the limitations of in silico work and are applying it responsibly.
- We will communicate our project in a way that balances hope and realism, echoing the clinicians' concerns about patient expectations and safety.
In short, the interviews reinforced that our idea is scientifically meaningful and clinically relevant, but also reminded us to frame it carefully, validate it iteratively, and stay aware of broader medical and ethical considerations.
