The root mistake, IMO, was not immediately noticing the obvious idea of classifying transsexuality as a value rather than a belief.

I'm not sure that "value" is the best way to think about transgender/transsexuality either. Standard decision theory only allows values and beliefs to influence decisions, but human behavior seems to be influenced strongly by identity, and transgender/transsexuality is usually described as identifying with a gender/sex other than the conventional gender/sex.

How to translate identity-based decision making into values and/or beliefs seems non-trivial, and can perhaps be compared to the problem of translating anticipated-reward type decision making into preferences over states of the world or over math. It's not even clear that such translation should be done.

If not, there are least two choices: one, just stick with our human preferences / decision process, or two, discard the old preferences / decision process, and generate fresh preferences from other applicable intuitions, as steven0461 recently suggested.