There's quite an amount of literature on what brain structures are correlated with what self-ID'd gender and sexual orientation. There are some in the collection I have:
https://dl.dropboxusercontent.com/u/14087916/transMedicalStudies.txt You'll have to read through the titles and/or abstracts.
So what does that mean for the biology viewpoint that seems to suggest to alter the mind to remove it? Probably not much, since whoever uses that argument surely hasn't put much thought into it and probably just found bible quotes that support their gay-hate. They certainly didn't think about how it could have once been a useful trait.
I don't think it's possible for a biological viewpoint to suggest altering the mind to remove it (or to suggest any action at all). That's a lot of motivation being tacked on to a science based on experimentation and observation. The motivation to "do something about the gays" isn't driven by knowing that it's possible, but by the assumption that it is necessary to do something about LGBTQ people (in order to accomplish moralist/religious/patriarchal/Nazi/arbitrary goal based an falsehoods/who-cares bullshit but this is where homophobia falls apart and shows itself as an irrational mess).
So if groups with gay animals were actually more successful to the point that it got adapted into the gene pool *, it must've been even before primates evolved and they just kept the trait and passed it on when they evolved to humans. But it's essentially useless today.
Well, if you want to examine modern use value or labour value: I reckon gay people are likely to adopt to start a family, so socially speaking, it's a service being performed. However, to start, I don't think variable sexuality is a negative trait at all, since love-making is a highly social activity that has a lot of health benefits (physical, emotional, mental leading to a lot of endorphin, stimulation, de-stressing, etc).
Besides the labour (is sthis what you mean? -- by what measure do you determine usefulneess) value: as Lanth alluded to, the biological factors that produce variable sexuality are necessary to produce normalized sexuality. If we look at "sexual attraction" from a more algorithmic perspective, we might assume that it is optimized to help individuals mate and reproduce. In that case, it is necessary for the algorithm to be lax enough to allow individuals with a wide variety of different phenotypes to be attracted to one another and attempt mating in order to preserve the qualities of a large and diverse gene pool. In this way, it's basically impossible for there not to be some amount of LGBTQ people.
That said, it is entirely possible that in the future the algorithm will result in a gender nihilist society in which attraction is variable and arbitrary (as in "who you like is who you like") because it is possible for us to be an example of a species whose intelligence and social engineering culminates into a society in which reproductive labour is much less associated with specific combinations of reproductive organs between mutually-attracted individuals. In other words, the reproductive labour that's commonly associated now with straight couples may become less associated together because of alternatives in breeding solutions.
In other words, assuming there is a "gay gene," it's overall consequences can be dramatically positive for a species, even for one such as ours.
(Also I feel like there's this underlying assumption that all gay couples or LGBTQ couples don't want children or to reproduce, which is proven false easily.)
edit: on one hand I have a longer ring finger and on the other I have a shorter one. I guess it checks out since I'm bi :') a y y