Faithful presentation of evidence

Figures and datasets must accurately represent the observations. Authors must not fabricate, selectively remove or alter evidence to support a preferred conclusion. Record acquisition and processing methods so that readers and reviewers can understand how the displayed result was produced.

Research images

Retain original files and acquisition metadata. Apply legitimate adjustments consistently to the relevant image and controls, and disclose nonlinear processing or other transformations that affect interpretation. Do not clone, erase, move or selectively enhance bands, cells or other features. Clearly mark boundaries when non-adjacent image regions or lanes are assembled and explain the arrangement in the legend.

For gels and blots, provide uncropped source images with lane identity and the link to the published panel. For microscopy and other imaging, report scale, sampling and the basis for selecting representative fields. Reuse of a control or image across panels must be scientifically justified and explicitly identified. The Nature Medicine image-integrity guidance offers further practical examples; HNCBT applies its own requirements described here.

Quantitative data and analysis

State the number of independent samples and the meaning of each plotted point and error bar. Report exclusions and missing observations with reasons, distinguishing prespecified rules from later decisions. Provide source data sufficient to connect the figures, tables and statistical results. Correct accidental duplication or labeling errors before submission and notify the editor if an error is discovered later.

Do not change outcomes, labels or analysis choices without disclosure in order to produce significance. Computational processing must preserve the provenance of the input and output records. Synthetic or simulated data must be identified as such and must not be presented as patient or experimental observations.

Editorial requests and concerns

Editors may request originals, source data, processing details or an explanation of a suspected inconsistency. Authors should supply these through a route appropriate to their confidentiality and access restrictions. The inability to substantiate an essential result may halt assessment. Concerns affecting a published article are addressed through Publication Ethics and Corrections & Retractions.