Using Spreadsheets In Bioinformatics Can Corrupt Data, Changing Gene Names Into Dates

from the careful,-now dept

A few years back, people were rather disturbed to find out about the famous Excel bug, whereby the multiplication of two numbers in Microsoft's spreadsheet gave the wrong number. It turns out there are other circumstances in which Excel (and, to be fair, presumably other spreadsheets) can give incorrect results, but they are unlikely to be encountered in typical everyday tasks. However, in the specialized world of bioinformatics, which uses computers to analyze data about genes and related areas, careless use of spreadsheets can throw up a significant numbers of errors, as this paper in BMC Bioinformatics explains:

Use of one of the research community's most valuable and extensively applied tools for manipulation of genomic data can introduce erroneous names. A default date conversion feature in Excel (Microsoft Corp., Redmond, WA) was altering gene names that it considered to look like dates. For example, the tumor suppressor DEC1 [Deleted in Esophageal Cancer 1] was being converted to '1-DEC.'

Here we have the interesting interaction of two very different fields, where the name of a gene involved in esophageal cancer, DEC1, was interpreted by Excel to mean the date, 1 December. As the paper points out, these kinds of substitution errors are already to be found in key public databases:

DEC1, a possible target for cancer therapy, was incorrectly rendered, and it could potentially be missed in downstream data analysis. The same type of error can infect, and propagate through, the major public data resources. For example, this type of error occurs several times in even the immaculately curated LocusLink database.

As that notes, a gene that might be relevant for treating cancer could well be missed because of this incorrect conversion to a date by Excel. Although it is unlikely that any serious harm has been caused by this -- yet -- it's a useful reminder of the dangers of depending a little too heavily on the results of software without checking for corruption of this kind.

Follow me @glynmoody on Twitter or identi.ca, and +glynmoody on Google+

Filed Under: bioinformatics, data, errors, genes, spreadsheets