Biomarker Research Workflow
An illustrative approach to organizing genomic data work, from source datasets to researcher review. This page describes a workflow, not a completed client study or clinical result.
Start With a Research Question
A biomarker project needs a defined question, suitable datasets, and a plan for evaluating candidate signals. Data preparation and analysis can support that work; their outputs require scientific interpretation.
For a sepsis research question, the scope might include comparing expression patterns across relevant study groups. Dataset selection, confounders, and the intended use of the findings must be agreed with the research team.
An Example Review Path
Data preparation
Document dataset provenance, inclusion criteria, transformations, and missing data before analysis.
Candidate analysis
Explore relationships in the data using methods selected for the research question and study design.
Review preparation
Package candidate signals, assumptions, and analysis code so a researcher can examine and reproduce the work.
Researcher interpretation
Review limitations, alternative explanations, and the independent evidence needed before further study.
Evidence Before an Outcome Claim
A candidate signal is not a validated biomarker. Claims about discovery speed, cost savings, clinical utility, or trial status require study-specific evidence and appropriate review.
For a published example of Michael Edwards's research work, see Crowdsourcing temporal transcriptomic coronavirus host infection data. That paper concerns its own research question and does not establish outcomes for a BioInfoAI client engagement.
Scope Your Research Workflow
Discuss your research question, data sources, and review requirements with the team.