R&D Experiment Tracking Workflow Checklist Template

• State the hypothesis — specifically and in falsifiable form; “Compound X will inhibit enzyme Y activity by more than 50% at concentrations above 10 nM” is testable; “Compound X will affect enzyme Y” is not
• Identify the independent variable — the variable being deliberately manipulated; what the researcher is changing
• Identify the dependent variable(s) — what is being measured as the outcome; must be quantifiable
• Identify confounding variables — factors that could affect the dependent variable other than the independent variable; controlled for in the design
• Ground the hypothesis in the literature — what is the evidence base that makes this hypothesis plausible? Literature cited in the ELN entry

The experiment should be designed to falsify the hypothesis, not confirm it. An experimental design that can only produce results consistent with the hypothesis — regardless of what happens — is not a scientific experiment.

• Design the experimental conditions — how many groups or conditions? What is the control condition?
• Include a positive control — a condition known to produce the expected positive result; confirms the assay or measurement system is working
• Include a negative control — a condition known not to produce the measured response; confirms absence of contamination or false positive
• Calculate the required sample size — using power analysis; based on expected effect size, desired statistical power (typically 80%), and significance threshold (α = 0.05)
• Consider blinding — blinding the experimenter to treatment allocation during measurement prevents unconscious bias
• Pre-specify the analysis plan — what statistical tests will be used? What is the primary outcome? Pre-specification prevents p-hacking

• Write the full protocol — before execution begins; in the ELN or lab notebook; specific enough that a colleague of equivalent skill could reproduce the experiment
• Include all materials — reagents with lot numbers, concentrations, and sources; equipment with models and settings; cell lines or biological materials with passage numbers and provenance
• Include all methods — step-by-step; timing; conditions (temperature, pH, humidity); the protocol is reproducible only if every condition is specified
• Version the protocol — if the protocol is revised before execution, the revision is documented with the reason; the executed version is identified in the experiment record

The lab notebook rule: if it wasn’t written down, it didn’t happen. Contemporaneous recording — writing observations at the time they are made — is the foundational data integrity practice in laboratory research. Reconstructed records are not original records.

• Follow the protocol as written — do not deviate without documenting the deviation; the executed experiment is what was done, not what was intended
• Record all conditions in real time — date and time of each step; equipment readings; environmental conditions; operator; material lot numbers used
• Record all observations — including unexpected observations; the observation that does not fit the hypothesis is as scientifically valuable as the one that does
• Document all deviations from protocol — what happened, when, why, and what impact it is assessed to have on the validity of the results
• Photograph or image capture where applicable — gel images, microscopy images, equipment readings; linked to the ELN entry; original unprocessed files retained

• Use the pre-specified analysis plan — analyse the primary outcome using the pre-specified method; any deviations documented and justified
• Process raw data transparently — all processing steps documented (normalisation, background subtraction, exclusion criteria); the processing workflow is reproducible from the raw data
• Handle outliers per pre-specified policy — not based on whether they change the conclusion; Grubbs’ test or equivalent if formal outlier testing is used
• Apply the statistical test — verify assumptions (normality, homogeneity of variance) before parametric tests; use non-parametric alternatives if assumptions are not met
• Retain the analysis code — if computational analysis is used (R, Python, etc.); the analysis code is part of the experiment record

• Interpret the results against the hypothesis — is the hypothesis supported, partially supported, or refuted? The conclusion follows from the data, not from the desired outcome
• Assess alternative explanations — what other explanations could account for the observed results? How do the controls address these alternatives?
• Identify limitations — sample size adequacy, measurement precision, potential confounds not fully controlled; honest acknowledgement of limitations is a mark of scientific rigour
• Write the experiment summary — in the ELN: hypothesis, design, protocol version, results, interpretation, and next steps

• Archive all experiment records — protocol (executed version), raw data, processed data, analysis code, results summary; in FAIR-compliant storage; with sufficient metadata for future retrieval
• Retain for the required period — minimum 5–10 years after publication for most research; longer for regulated research (clinical trials: 25 years; GLP studies: 10–15 years)
• Make data available for peer review — upon publication or on request; data sharing per the data management plan and funder requirements