TCB Ethics Guidelines

Adapted from the NeurIPS Ethics Guidelines. For additional guidelines on ethical publishing, see our editorial policies.

Introduction

Computational biology research can have significant real-world impact on medicine, agriculture, environmental science, and human health. As this impact grows, so does the potential for harm. Issues with data privacy (especially genomic data), algorithmic bias in clinical applications, dual-use concerns in synthetic biology, and the misuse of computational predictions have been documented in the literature.

Researchers in computational biology should consider not only the potential benefits but also the potential negative impacts of their work, and adopt measures that enable positive trajectories while mitigating risk of harm. We expect authors to discuss ethical and societal consequences in their papers, while avoiding excessive speculation.

This document is intended for both authors and reviewers to establish common ground about the ethics guidelines. In rare situations, TCB reserves the right to reject submissions that have violated key ethical principles.

There are two aspects of ethics to consider: potential negative societal impacts (Section 2) and general ethical conduct (Section 3).

Potential for Negative Societal Impact

Submissions to TCB are expected, when applicable, to include a discussion about potential negative societal impacts of the proposed research or application. Whenever these are identified, submissions should also include a discussion of how these risks can be mitigated.

Particular concerns in computational biology include:

Dual-use risks. Could the method be used to engineer pathogens, design bioweapons, or otherwise be applied in ways that could cause mass harm?
Genomic privacy. Does the research involve human genomic data? Could results or released tools allow re-identification of individuals? Have appropriate consent and anonymization procedures been followed?
Bias in clinical or diagnostic applications. If the method is intended for clinical use, does it perform equitably across populations? Has it been evaluated for potential biases related to ancestry, sex, age, or other demographic factors?
Raise human rights concerns. Could the technology be used to discriminate or otherwise negatively impact people, including impacts on healthcare access or insurance?
Environmental impact. Does the computational work involve large-scale training or simulation that has significant energy or environmental costs? If so, is this cost justified?
Deceive people in ways that cause harm. Could computational predictions (e.g., of drug efficacy or disease risk) be misleading if misapplied or misunderstood outside of the scientific community?

General Ethical Conduct

Submissions must adhere to ethical standards for responsible research practice.

If the research uses human-derived data (including genomic, clinical, or imaging data), consider whether that data might:

Contain personally identifiable or sensitive information. Has consent been obtained? Could individuals be re-identified from the data or results?
Contain information that could be deduced without consent. For example, inferred disease status, ancestry, or relatedness.
Encode or exacerbate bias against people of a certain gender, race, ancestry, or other protected characteristic.
Require ethical board review. Studies involving human subjects or patient data are expected to have been reviewed and approved by a relevant IRB or ethics committee.
Have been discredited or retracted by the creators. Such datasets should not be used.

Additional data-related considerations include:

Consent to use or share the data. Have you obtained permission from data owners? If not, explain why this is ethically appropriate.
Domain-specific considerations when working with high-risk groups, such as patients, minors, or vulnerable populations.
Compliance with GDPR, HIPAA, and other data regulations, as applicable to the jurisdiction and type of data used.

Final Remarks

We expect TCB submissions to include discussion about potential harms, malicious use, and other potential ethical concerns arising from the use of the proposed approach or application, including biology-specific risks such as dual use, genomic privacy, and clinical bias. Submissions will be evaluated on the depth of such ethical reflections.

References

[1] J. Whittlestone, R. Nyrup, A. Alexandrova, K. Dihal, and S. Cave. (2019) Ethical and societal implications of algorithms, data, and artificial intelligence: a roadmap for research. London: Nuffield Foundation.

[2] B. Hecht et al. (2018) It's Time to Do Something: Mitigating the Negative Impacts of Computing Through a Change to the Peer Review Process. ACM Future of Computing Blog.