Here's what I learned from my experience:
Before the talk
Start making slides early.
Because so much of my data involves number crunching and tables and graphs, it was getting really tedious sifting through folder and folder and list after list of files. So eventually I started keeping a running powerpoint of my work and using the comments box for descriptions. Maybe people already do this, but it didn't occur to me to start curating my work this way until I was swimming in data tables. It turned out to be a huge time saver because when it came time to put together my formal talk, I had already made figures for most of my main points.
Give a practice talk.
This serves the dual purpose of making sure you don't leave the presentation making to the last minute and also gives you an opportunity to get feedback on your analysis and presentation style. I gave a formal lab meeting one week before my talk, which helped other members of the lab get up to date on my work, and also gave me a chance to get some input regarding other analyses I might want to implement. Then the day before my presentation, I gave a slightly less formal presentation to my friends (some non-scientists, and some scientists outside of my department) to make sure I wasn't using too much human-specific jargon or glossing over essential background information. Both of these experiences were really useful, and I got some very good tips from both audiences. Lastly, giving a practice talk lets you ensure that you can cover all of your material in the allotted time.
The day of the talk
Get to the presentation venue early.
Our department talks are always in the same room, so this was less of an issue. However, it did allow me to set up on time and avoid the awkward five minutes when everyone watches the presenter fumble with the projector and display settings. Getting there early also let me realize that I had left my laser pointer and adaptor in lab, so I had enough time to run back and grab those. Also, it just helps to get used to the room, so that you can focus on the data you're going to present, rather than distractions like the incessantly flickering fluorescent bulb near the door.
Once is not enough.
If a particular topic is central to understanding your research, it is not enough to mention it once. You need to constantly remind your audience about this fact. In my case, I created a navigation slide to give people some idea of where I was in my talk, and after covering each point, went back to a list of research questions and checked off the things I had already answered. Even so, there will still be people who ask you to explain that topic yet again. This is just the reality of presenting to a room full of overspecialized individuals that dedicate only these thirty minutes of their time to your project. Remind people often what your research goal is and why it matters (despite what the best cynics might say, yes, your research matters).
Leave time for questions.
This can be hard if you have a lot of data to cover. But a deadline is a deadline. Think about the pieces of data that are most essential to telling your story. Other connecting bits may be interesting, but not essential. Leave these out. If it's really a missing link, your audience will ask you about it, and then you'll have the added bonus of being able to answer their question!
It's also a good idea to anticipate some of the questions you might be getting, especially regarding what you plan to do next in your experiments.
Be able to defend your answer.
There is nothing more embarrassing than when the presenter is obviously thrown by a question and rambles their way through it. The audience would gratefully prefer a succinct "That's a great question. I'm not familiar with any data that explains that." Alternatively, there's the even shorter, "I don't know, but I'll look into it." Or the even shorter, "I don't know." Gasp.
After the talk
Be gracious to the audience for their time and attention. Even if half the audience is there for the free coffee and bagels, everyone in the room should have at least some marginal interest in science. So give them a scientific presentation. Make it visually appealing. Speak clearly and look at your audience, not your slides. Also, be gracious to the people brave enough to ask questions. Lastly, make sure you have an acknowledgements slide. I was really humbled when I saw just how many people put in time and effort along with me to answer "my" research question. Thank your advisor, your lab, your collaborators, your funding source, your audience, and above all else, the patients you've never met. Without their consent, you'd have nothing.
It's no small task condensing a year's worth of research into thirty minutes. Once your laboratory-house is in order, and your PI gives you their two thumbs-up, treat yourself...