A higher gene-expression signal in mixed tissue has at least two possible explanations. Cells may be expressing the gene more strongly, or the specimen may contain more cells that already express it. Without separating those possibilities, a tissue-level difference can be mistaken for a change within cells.
I reanalysed public transcriptomic data related to diabetic retinopathy, focusing on TBXAS1. The work examines changes in gene activity and changes in the cellular composition of a specimen, then asks what needs to be tested next.
More cells do not mean more independent patients
Transcriptomic data measure the activity of many genes together. Whole-tissue and single-cell measurements present that activity differently. When several cells come from one patient, increasing the cell count does not create an equivalent number of independent patients.
I separated patient- or donor-level comparisons from analyses locating a signal within tissue. Rather than pooling different specimen types as interchangeable evidence, I assigned each source a role in answering the research question.
Does the interpretation depend on the calculation choices?
Comparisons require preprocessing to account for differences such as measurement scale. Interpretation can depend on those choices or on whether a particular specimen is included. I tested alternative conditions and prepared the validation report and manuscript around those checks.
My question goes beyond naming an eye-catching signal as a disease marker or treatment target. Which cells contribute to it, and what else would have to be measured before calling it regulation within the same cells? Those distinctions help define a follow-up experiment. The reanalysis does not establish clinical validity or a mechanism.
Preparing a Technical Brief
The record dated 5 September 2026 places the manuscript under review at Molecular Vision. I agreed to the editor’s proposed Technical Brief route and prepared a revised manuscript to use if requested. At that point, I had asked whether the existing files would be used or a new upload was needed and was awaiting guidance; the reserve files had not been resubmitted.
Numerical results and clinical interpretation remain unpublished. Once there is a live version of record and an active DOI, I plan to explain the sources and comparisons used to separate cellular composition from molecular regulation alongside the findings.