Excel and Genomics was a recipe for a disaster. Again, for home/office users, something like Wordpad with small tables, such as the ones from Ted works for a 99% of the cases.

Everything else should never interact with a crap like Excel. Ever. Use dedicated, scientific journals for it if you are a self-called professional. Reusing a ledger sheet for serious stuff will just make companies lose billions with faulty research and papers.

Access? Nice toy, but even Sqlite3 as a damn engine can be more powerful and straightforward once you stick any RAD UI, from Lazarus to something small in C#, which is possible in AOT. Heck, even if you have Java, by decoupling the backend you are safe to put even a TCL/Tk GUI, even it if's looks 'horrible' by default, but for serious use, you would love the default "industrial" design. Or just use the Win32 TTK lookalike which would look well on almost any PC.

Ah, yes, reports and the like. Use a damn library to print to PDF, there are zillions. Or, just better: once you get the result in text, just paste the results back to Ted/Writer/Word and then you are free to compose the document in any style you like. Then, well, 'print' to PDF, or PS->PDF.

It might be more complex, yes. But your data won't be mangled in the process. The results would be in a separate text file, outside of any faulty spreadsheet or wannabe database. That output should be pristine and backed up ASAP. As it's text, it's parseable from nearly every machine from 2002. Even more if it's plain ASCII, for sure a thing with science done in English and plain numbers. No localized MSOffice bullshit for formulae, no more numbers parsed as dates, ever.