To start with, you might try a program that can predict the arrangement of transmembrane proteins based on sequence. The program, described in more detail than most people will want to see here, uses rules deduced by biologists, and was then fine-tuned with an artificial intelligence algorithm, training the program using transmembrane proteins with known orientations. There is a similar program that can predict where in a cell a protein will end up.
Then you can try other predictions that can be made based solely on sequence by searching online for other predictor tools.
(2) See how little a well-conserved protein has changed through evolution. Let's look at the current versions of beta-tubulin from yeast and human, and see how similar they are. Try this: copy the text of the yeast beta-tubulin sequence from here (from MREI... to the end of the protein's sequence), then paste it into the Enter Query Sequence box here, under Choose Search Set, for Database choose non-redundant protein sequences (nr) and next to Organism, type "human" and then select human from the dropdown menu that appears ('human' or 'humans'). Then click BLAST at the bottom. The page will automatically update for seconds or minutes, depending on how busy servers are. When it's done, you'll see the results in a tab marked Alignments. You'll see the sequence you queried (the yeast beta-tubulin) and the subject sequence – the closest protein sequence that it could find among all known human proteins. In between is a list of identical amino acids, and + signs for similar amino acids (similar based on charge, etc). Tubulin, actin, and histones are remarkably well conserved proteins – they've had very few changes across hundreds of millions of years of evolution. If you try the exercise with other kinds of proteins, you'll see that only parts of them are well conserved across diverse organisms, or that some don't exist in certain organisms.
More about Bioinformatics at Wikipedia.
(image: a 7-transmembrane protein from Wikimedia Commons)
















