Auto-correction features have been much maligned for resulting in misleading or incoherent text messages and emails, but they can also mar important scientific research.
Australian researchers recently published an analysis that found a feature in the spreadsheet program Microsoft Excel has created errors in about 20% of genetics papers published in leading scientific journals such as Nature and Science.
It happens when the symbols used to denote genes in scientific literature are misinterpreted as dates or other numbers. For example, the gene Septin 2 is usually referred to as SEPT2. Excel automatically assumes the person inputting the number means the date of Sept. 2 and stores that in the field.
This cannot be turned off, so to avoid the mistake, every field must be changed separately. This, not surprisingly, is not always accomplished.
These are far from the first cases of computer data-entry errors that have led to possibly flawed research, and the analysis notes that the gene problem was first noticed more than 10 years ago.
The authors of the new paper reviewed about 3,600 genetics papers by looking at attached supplementary files that contain the list of genes involved in the research.
They suggest “researchers, reviewers, editorial staff and database curators remain vigilant” to avoid any significant errors.
