[chromatin]

Perversely, this is almost the exact opposite of how to solve the problem.

Speaking from experience, the real problem with scientific data (genomics, in particular) is when it is exchanged in plain-text format and then loaded IN to Excel. Automatic type inference is what fails.

Data already in Excel format can be created/modified/exchanged without unintentional conversion if the column type is correctly specified as "TEXT".

[rbanffy]

True, but how do you trust something that started as a CSV and was digested into Excel?

[chromatin]

Unless you've done it yourself carefully, you can't