I am sure others can give you more elaborated answers, but I just start: yes, I think if you align several of the "other" proteins with yours, you can construct a phylogenetic tree (I assume you have the DNA sequences of all of them). Then, you can pick any reliable existing phylogeny of bacteria and "root" your tree. Once this is done, you might get a better picture of where your protein branched off. If it branches of somewhere further up in the tree, then this is evidence that your protein is a more recent version. If your protein branch connects to the tree at the very bottom, this could be an indication that it is indeed a very old relationship (whether you wanna call it an "ancestor" is not quite clear to me, could still be either way). But beware that phylogeny programs might sometimes be fooled by fast rates of evolution and then resort to placing all fast evolving sequences at the very bottom of the tree (long branch attraction).

Disclaimer: haven't worked for a while with phylogenies.

Dominik is quite right. In particular long branch attraction give misleading phylogenies if any of the sequences have long branches (lots of substitutions) and are joined by relatively few substitutions (the so-called Felsenstein Zone) . Algorithms will generally group these long branches that are in fact not closely related. This is where using an optimal maximum likelihood model of sequence evolution is really critical.

Astonishly perhaps, phylogenetics can provide information on the branching pattern and amount of evolution that has happened throughout the evolutionary process. However, the tree that is inferred is, in all cases, an unrooted tree. Finding the root adds such a degree of mathematical complexity that computers are not able to evaluate the vast number of topologies and the placement of the root on these trees.

So we root a tree in order to find the ancestral state using data that is from other sources. Sometimes that is biogeographical or morphological or perhaps from some outside molecular information that gives clues to who is the closet to the so-called 'most recent common ancestor". This requires some general knowledge of the biology of the organisms and is ofter ignored with the result that the root is misleadingly placed.

You already got good answers above. I'd just like to add one technical detail, that no organism or protein found today is really "ancestral" to another. Chimpanzees for example are not "ancestors" of humans, we did share a recent (maybe 8 - 14 million years ago) common ancestor with chimpanzees and gorillas, but modern chimpanzees and gorillas are not our ancestors.

An amino acid sequence identity level of 30% is reasonably high, given that with 20 amino acids, the odds of matching at a given site are closer to 5% at random. But it is more like comparing humans to fungi, than humans to other mammals or humans to insects. Many mammal amino acid sequences are more than 80% identical to their insect counterparts. And in fact many mammal amino acid sequences are more than 60% identical to their fungal counterparts.

Indeed Brian. The MRCA (most recent common ancestor) is a construction from morphological, fossil and molecular data we have on hand today. It is a theoretical ancestor, but well founded if our data is sufficient and from different sources (not one point estimate). From this we can construct a hypothetical ancestor to gain the direction of evolution over millennia. We cannot know the past (epistemologically) but we have wonderfully informative data that gives us a reasonable robust estimate of the characteristics that the ancestor must have possessed.