If we assume that Excel is an unavoidable evil (and for most this is unfortunately a credible assumption)...
When importing your delimited data into Excel, be sure to set the column type appropriately. The default for all columns is 'General' - this will lead to the inevitable auto-correct mangling. Gene name columns should be set to 'Text', then the septins etc will be left unmolested.
Once mangling has occurred, correction needs to be approached carefully - the main issue with what Excel does here is that it changes the underlying data, rather than only changing things in the display layer. This means your gene names are turned into floats (which is Excel's underlying representation of dates). Of course, Excel being Excel, the float used is different depending on the platform: https://support.microsoft.com/en-us/kb/214330 (also some versions of Excel think 1900 was a leap year, others know better - this is a very deep rabbit hole...). You could conceivably write some code to handle the conversion of dates back to gene names, but it would have to consider these caveats and edge cases veeeerrrry carefully.
All in all, I agree with Michael (although accept that this is not always possible).
Please read this paper before making any efforts and then decide for yourself!