Library-size normalization is confounded with disease stage in human liver transcriptomes: code, pre-registration, survey records and results

Code, pre-registration with thirteen dated amendments, literature survey records, result files, figures and machine verifier for the manuscript of the same name.

The study reanalyzes five public bulk liver RNA sequencing cohorts of fatty liver disease, 610 biopsies in total, and shows that the share of the sequencing library held by the most abundant transcripts covaries with histological disease stage. A size factor based on total counts is therefore confounded with the contrast of interest. The proportion of the expressed transcriptome appearing to move in one direction with disease spans 57.2 percentage points across the five cohorts under counts per million and 10.1 points under the median of ratios, with nothing changed between the two results except the size factor. A survey of 331 papers reusing these accessions found that about half do not state which normalization they applied.

Input data are public and are not redistributed; the accessions and the exact files required are listed in README.md, together with the conventions needed to reproduce each number. The composition arm is reported as void by its own pre-registered positive control and is deposited in full, because a failed control is a result.