I read this article today (poorly written and not very informative. Don't bother reading).
Tchantchaleishvili V, Schmitto JD. Preparing a scientific manuscript in Linux: Today's possibilities and limitations. BMC research notes. 2011 Oct;4(1):434+. Available from:

.
The table at the very end summarizes the format in which top biological sciences journals prefer to receive submissions.
- New England Journal of Medicine(34) PDF*, DOC, WPD, TXT, RTF
- Cell(37) DOC, RTF, TXT
- The Lancet(38) DOC
- JAMA(39) DOC, WPD
- The Journal of Biological Chemistry(35) PDF***
- Circulation(40) DOC, WPD
- ===
- Nature(29) DOC, TEX**
- Science(30) DOC, TEX**, RTF*
- PloS ONE(31) DOC, TEX, RTF
- PNAS(32) DOC, RTF, TEX
- BMC Journals (33) DOC, RTF, TEX
[* Some restrictions apply]
[** TEX files must be accompanied by a PDF version of the same text for visual reference 19 ]
[*** Although manuscripts must be submitted in PDF format, Microsoft Word is recommended to prepare the manuscript text]
Look at the sheer proliferation of M$hit formats on that list. The 1st preference is always doc. Wherever PDF is accepted, M$hit is recommended to make that PDF.
Even if you want to change *they* will not let you.
Ugh.
Am I in the wrong field or what? Nevermind content, even submission guidelines are so proprietary, backwards-thinking and insular.