FSL is the FMRIB’s Software Library, one of the most widely used neuroimaging analysis packages in the world, and is developed and maintained by the members of the Analysis Group in Oxford. If you visit this ipython notebook you can see my calculations for how many emails were sent on their mailing list in April 2014: a mind blowing 871!
Freesurfer is another extremely popular tool developed by the Laboratory for Computational Neuroimaging at Massachusetts General Hospital in Boston. I haven’t calculated all the emails from their mailing list, but there were 41 on a representative day in December (yesterday) so I think I can safely conclude that those researchers aren’t sitting about doing nothing either.
Although anyone can reply to emails on the list, there are a small number of developers who will most likely to reply to your email, and the number of answers that they provide on a (literally) daily basis is awe inspiring.
They want you to use their tools, and the very first email to the FSL list said it all:
If you think about it, there’s a really strong chance that – even in just one month – someone asked a question that’s relevant to you:
- Are you learning a new neuroimaging analysis technique for the first time?
- So have thousands of others before you!
- Are you unsure about your statistical analysis plan?
- So were thousands of others before you!
- Are you confused by an error message?
- Thousands of others before you have stared at the words “child process exited abnormally” with absolutely no idea what went wrong and I suspect many more will in the future.
My point, as I’m sure you’re getting, is that there are very few new questions. So rather than just jump in and ask someone, it’s your job as a new scientist to learn from those who have come before you.
I would put the most helpful emails broadly into two categories: understanding error messages, and understanding statistical models. If you’re getting an error message when you run FSL, you probably aren’t the first person to see it. It’s almost guaranteed that someone else has had the same difficulty as you, has asked the question before, and has received help. And while your statistical model will be specific to your data set, there are almost certainly analogous models that have been used in the past.
All the information is in those mailing list archives, just waiting for you to help yourself.
So, here’s what you have to do:
- The FSL and Freesurfer mailing lists are very difficult to search, so step one is to subscribe to the lists and create your own archive. I use gmail and find that searching the 7 years of archives (I’ve been at this for a while) in my account makes finding appropriate emails much easier.
- Read the subject lines of all the emails. I’m in the UK and tend to wake up to a lot of emails from the USA, so I have a little zen moment every morning where I archive all the FSL and Freesurfer emails from my inbox. Given that I’ve been reading the emails for a long time I tend not to read many, but when I see something about a new feature, or a question that seems like something that might be relevant then I go to step 3…
- Read the emails that might be interesting and bookmark those that are useful. If you see someone asking about a problem with dtifit and you know you’re going to be conducting DTI analyses then read that email. Even if it isn’t a question you have now, there’s a good chance it’ll be relevant in the future. If it really makes sense then bookmark it so that you can easily find it in the future. You can do this in gmail either with clever use of labels etc, or I often just email it to myself with a bunch of words that I’m likely to try to type into the search bar when I look for it in the future.
- Read all emails about statistical tests that you don’t understand and try to answer (in your head) each person’s question before you read the expert’s answer. These are prime learning opportunities, and there are profound similarities across all tests. Use these worked examples to hone your understanding. Talk through ones you don’t understand with friends and colleagues. Create a database of examples that are useful and refer to them often.
If you do have to send an email to the list make sure that you’ve done your due diligence: search through your archives, and the general mailing list archive for your particular error message and make sure you completely understand all the statistical examples on the appropriate help pages, including, but not limited to: these FSL GLM examples, Jeanette Mumford’s advice on demeaning covariates and these Freesurfer GLM examples.
Only once you’ve convinced yourself that a simple LMGTFY search won’t give you the answers you need, then you can write your email to the list. I don’t want to paralyse you with fear, but thousands of people will read that email: make sure you look good while you’re asking your question!
- Make sure your email is polite – these people are busy and not paid anywhere near enough for their jobs. A little courtesy goes a long way.
- Explain concisely what you want to do and where the problem is. Don’t put in more detail than you need, but do make sure that someone can follow your question easily.
- If you’re asking for help with an error message include the full error screen printout and the command you typed before it appeared!
- If you have a problem with viewing an image then take a screen shot and annotate it to point out exactly what you’re worried about.
- Include in your post what you’ve tried and what resources you’ve exhausted. If there’s an old email on the list that you found but didn’t understand, cite it. Show the readers that you’ve done some work towards answering your own question.
- Give your email a sensible subject line so that others can learn from it. Just think of all those times that you’ve read others’ questions and consider this your way of giving back to the community!
- Before you send the email, do one last search in the archives. If you’ve gone through all these steps to make your question really clear, you may find that you can search more appropriately and those answers may be there ready and waiting!
It takes a while, but if you keep at these lists for a while you’ll start to see what I’ve already told you: there are very few new questions it’s just a case of knowing what to search for.
If I were running a start up I’d strongly consider hiring people who had successfully completed a PhD in any discipline because I know that their Googling skills are among the best the in the world. It may not seem like “doing science” but figuring out how to find information from a variety of sources is fundamental to your success as a critical thinker. Research isn’t just standing on the shoulders of giants, it’s finding your way up to there in the first place!
After a while, you’ll be able to answer some of the questions yourself, and you can either transfer that understanding back to the email list, or at least be able to help your colleagues out.
Good luck, you got this.