[Oct 7] Click here for updated information on the Conference Venue
With the 10Kin1Day connectome workshop we want to bring connectomics to you in the easiest way possible!
In this 3-day event we aim to bring together a lot of groups that work on DTI data analysis across the world, in particular those that are working within the fantastic ENIGMA consortium, and start doing some real hands-on connectomics! So if you have DTI data lying around and want to have that converted into connectome maps + you are interested in hands-on instructions on how to start your own connectome project, this is the workshop for you.
The plan is for everyone to bring his/her DTI and T1 data to the workshop (large datasets encouraged) and then together run as a group connectome reconstruction and analysis. We reserved computational power in the cloud on the Dutch national supercomputer (having a total of almost 41,000 computer cores, lots of memory, secure access) to run all of the data during the workshop.
Every one of the participants will get his/her own account on the supercomputer, so there is no need for sharing any data. Put differently: we really don't want your raw data. The goal is to run as many connectome maps as possible, and then share the outcome connectivity matrices with you for further analysis within your own lab and, if you wish, within the ENIGMA working groups. How cool would it be, if we can bring together the largest connectome dataset of the world??
10Kin1day is very generously sponsored by Neuroscience and Cognition Utrecht (NCU) of Utrecht University, the ENIGMA consortium (thanks Paul!), the Department of Psychiatry of University Medical Center Utrecht, and dutchconnectomlab.org (us). Thanks to this, we think of keeping the participation fee as low as 250 euro, which includes everything (housing for 2 nights, breakfast, coffee(!), lunch, afternoon tea, and some snacks during the day). We will even get you a nice diner in the evening. Oh, drinks at the bar are at your own expense (you can buy us a beer if you want), but if we hit the 10K we might go crazy...
Day-to-day programClick here to open the program (pdf)
VenueThe venue is a very nice conference center in the woods close to Utrecht, Netherlands (www.woudschoten.nl). The place is reachable very easily by direct train from Schiphol Airport towards Nijmegen, getting off at station Driebergen-Zeist (~45 min) + a short taxi ride (~10 min) or (~15 min) bus ride (bus 81 towards ‘woudenberg de poort’, getting off at the last stop). If you choose to take the bus it is 5-10 minutes walk from the bus stop to the conference center. You can also take a direct taxi from Schiphol for around 75 euros, but let us know so we can book this for you and get you a fixed price! We have pre-booked rooms in the conference center, so you do not need to book a hotel yourself!
The full address of the Conference center is:
Woudschoten Hotel & Conferentiecentrum
Woudenbergseweg 54, Zeist
Map of the Conference center and wider area
Map of the Conference center and the walk from nearest bus stop
Leave your info below if you're interested to join the event. We'll keep you posted about the program.
This event is organized by Martijn van den Heuvel (dutchconnectomelab.org), Paul Thompson, Neda Jahanshad, Marcel de Reus, Ingrid Romme and the ENIGMA consortium.
Wowie, can I ask some questions?
Should I bring my own connectome software?
Nope. We have that covered. We know that there are many ways do connectome reconstruction, but the goal of this event is not to compare different methods, but to start doing connectomics. We also want those people that want to start on connectomics, not only have those that already know how to do it. So, during the event we will use our custom-built connectomizer software package to convert your DTI data into connectomes. We made an easy-to-use connectomizer version specifically for this event, running eddy-current and movement correction, robust (RESTORE) tensor reconstruction, and streamline tractography, combined with the Desikan-Killiany atlas for region selection. Running this package will be super easy.
What should I bring?
For every one of your subjects you should bring the T1 scan, FreeSurfer output (you thus have to run FreeSurfer on every one of the subjects, you can bring manually corrected segmentations if you want), a DTI file (4D nii.gz file, containing the b0 and diffusion-weighted scans), and a bval and bvec file. So you should bring:
* FreeSurfer output (complete directory created using recon-all or equivalent steps)
* bval and bvec file
* the connectome spirit
* a laptop with MATLAB and the Statistics and Machine Learning Toolbox installed (there are free trial versions and student versions at mathworks.com)
Can you process 10,000 scans?
Yes we think we can (but nobody has tried this before). We are kindly sponsored by SURFsara (the organization managing the Dutch national supercomputer) and the Utrecht Bioinformatics Center (thanks Patrick!), with a lot of CPU's ready to go. However, to avoid creating a bottleneck by uploading 10,000 MRI datasets onto the cluster on 1 day, we plan to open up your account on the cluster 2 weeks before, so you can already upload the data to this secure location and have it ready to go when you come. We will send you your account credentials on time before the event. If you are not able to upload your data in advance, please contact us so we can find you a personalized option to upload the data.
Should I do something before the event?
A few things here and there, but the major steps (like the processing and first connectome analysis) will all be done during the event. Obviously, that's what the event is for. As mentioned above, we will be asking to upload your data to your account on the supercomputer before the event to avoid any upload bottlenecks (around 1-2 weeks before, details will follow) during the event. There is no-way we can upload all 10,000 datasets on the spot. We will provide detailed instructions on how to make this upload (provide your account information etcetera) + to do some pre-checks before the event on whether your data is in the correct order, whether all files are there, whether the bvals/bvecs fit with the DTI data, and so forth. Again, detailed descriptions will be provided, and we will open a ‘helpdesk’ and a ‘skype-hotline’ to help you, a few days before the event (announced on exact dates will be made).
This sounds complex?
No worries, we are easy, so our instructions will be easy as well. And, as we are lazy too, this won’t take too much of your time. If you don’t want to or you can’t do some of these pre-checks, it is also OK with us if you would send 2 or 3 test sets to us before the event (so complete datasets, for example by means of WeTransfer). We can then do all of the pre-flight checks for you, and/or give you feedback on whether your data is in the right format, order etcetera.
How important are these pre-flight checks on my data?
We think ‘very’. The thing is that the event is only 3 days, and we want to spend as much time as possible on running all of the data (so we want to have all of the sets started as soon as possible), and as much time as possible on the lectures/hands-on workshops on the connectome analysis. It would be a shame to waste time on boring things like uploading, fixing data formats, ordering of files etcetera. That’s why we want to have everything prepared before the event, so we are ready to run on all cylinders from the first minute on!
OH NO! I can't make it to Utrecht on 28-30 October. Can I still participate?
Absolutely! If you want, you will can still get your own account on the supercomputer . You can then upload the data there, and have it ready to go. During the event, someone of us (from the organizing committee) can then run the data for you, and you can download the end results. This does of course mean that you have to allow us to 'hit the button' for you during the event. We are happy to sign any types of 'agreements' to state that we won't do anything with the data, except for running the connectome software.
What is for me to keep?
Everything. All of the output (registered T1/b0 files, fiber clouds, matrices, etcetera) will remain yours to keep. You can download this data during or shortly after the event to your home/institute computer. The connectivity matrices will be directly usable for your own scientific project (e.g. your case-control study, your developmental study, your simulation study on controls, and so on). Reconstruction of connectome maps will be the same as for our own connectome projects (see dutchconnectomelab.org for our work) and (in our opinion) provide a good basis for scientific analyses and later publication. We will also do our utmost to prepare a version of the connectomizer binary for you to take home, made suitable to run on your own Linux machine or Mac laptop. If we run out of time, we will send this one to you later.
My dataset contains sensitive data, is there a possibility of having a data agreement?
Absolutely. Yes, please send us the forms and we will be happy to sign any document stating that we will by no means work on your data before, during, or after the event without your permission, and that we will under no circumstances publish about your data. Remember, you will be given your own secure account with your own username/password, so even if we would have wanted to, we wouldn't be able to access your data without your consent. If you don't delete your data when you’re done using the cluster, the Dutch supercomputer will do this automatically 2 weeks after the event. The only thing we want to promote is to share the connectivity matrices within the ENIGMA consortium (still with your permission, and all accompanied authorships etcetera), so the working groups can start their own individual projects on connectomics.
A very nice conference venue in the woods close to Utrecht, Netherlands. The place is reachable very easily by direct train from Schiphol Airport (~45 minutes) + a short taxi ride (~10 minutes). You can also take a direct taxi from Schiphol for around 75 euros, but let us know so we can book this for you and get you a fixed price!! Everything is located in one spot (hotel, conference venue, restaurant).
Schiphol is the national airport of the Netherlands, hosting flights of both the big airlines (KLM, United, Delta, etc.) as well as smaller ones (EasyJet, etc.). Typically, in particular within Europe, you can fly for ~250 euros return. The event takes places during a weekend (Friday-Sunday), so fares should be lowest.
Can I arrive on Thursday 27th?
Yep. Let us know and we can arrange 1 night additional stay. Cost of this will be ~100 euro including breakfast.
Can I stay until Monday 31st?
Yep. Let us know and we can arrange 1 night additional stay. Cost of this will be ~100 euros including breakfast.
What are we going to do during these 3 days, besides having fun?
The program will be as follows (tentative, we might move around something here and there): Day 1: welcome, instructions on the pipeline we are going to use (including a detailed explanation of all intermediate steps), hands-on workshop on running the software on a few example subjects, explanation of quality control steps, submitting all data on the cluster for massive number crunching
Day 2: breakfast, looking at progress, downloading first datasets, workshop on quality control (do my matrices look OK?), educational lectures and hands-on workshop on 'graph theory for connectomes' (applied to your own freshly baked connectome matrices), nice walk in the woods, dinner, drinks
Day 3: breakfast, educational lectures and hands-on workshop on 'group connectomics' to start and exercises on starting your own (and hopefully witin an ENIGMA working group) connectome study
I have a lot of disease MRI data. Is this for control datasets only?
Absolutely not! The event is PARTICULARLY suitable for starting your disease connectome study. So bring the patient and control data in one directory and run them. Even more: the ENIGMA consortium has many working groups on different brain disorders and it would be fantastic if we could start connectome projects in every one of those groups. On day 3 we will discuss ways to compare and analyze disease connectome maps.
Is there an upper limit on datasets?
No way! Please bring those large datasets. We want to hit that 10K, so we absolutely encourage you to bring your entire dataset. Oh, and let us know when you have >1000 scans, so we can buy enough disk space :) Is there a lower limit on datasets?
No, not really. We want to keep the event open to anyone. However, given that we are limited to 50 seats, we are aiming at datasets of ~50-100 scans and more. This is roughly the minimal number of scans needed to do a MRI study (case-control, aging, etc.), and we really want that people can start working on their own connectome study when they return home.
I don't have that number of scans...
Well if you are still willing to come, drop us an email. The thing is that our grants cover the analysis of data (our goal), but we would be OK if you would be willing to carry the housing/food/coffee costs yourself. This will be around 500 euro on top of the 250 euro participation fee, so 750 euro for all 3 days (and yes this includes: housing/breakfast/lunch/coffee). Our grants cover up to 50 people.
What if more than 50 people want to join?
Likely, as this will be an awesome event. We really hate to say 'no' to people, but in that case we will need to shortlist people that are very serious about doing a connectome study (convince us by sending us an email explaining why you want to participate), people from the ENIGMA consortium and/or those that are willing to team up with ENIGMA.
The coolest consortium on MRI ever. 50,000 datasets, 200 groups, 100 countries, led by Paul Thompson from USC (LA, USA). The ENIGMA website is here: enigma.ini.usc.edu. Take a look, it's brilliant.
Why are you doing this?
Because we like people, and, above all, we like connectomics. In the last 2 years we have had a very cool and (as we dare to say successful) summer school on 'graph theory for brain networks' (utrechtsummerschool.nl) as part of the summer school courses of Utrecht University. We will continue with this of course, the next one will be in the summer of 2017. However, what we noticed is that many people would love to go one step back, and actually want to learn about how to process their raw DTI data into a connectome map. This without any hard-core coding, but just a ready-to-go way to process data. Moreover, in many labs, and in the ENIGMA consortium, there is a lot of DTI data available, but there is not always time to process this data into connectome maps. Therefore, we offered to take on this challenge, and help people to start doing connectomics in a fast and easy way. Our goal is thus certainly not to 'get all of the data and do our own project' but instead encourage others to start doing connectomics. This to catalyze connectomics, as well as collaborations between different groups (we really hope that the event will help many of the ENIGMA working groups).
In the end, we want to bring along people that are like-minded on connectomics, like to play with BIG DATA, and think of joining the ENIGMA consortium. On a serious note, we of course would love to see connectomics becoming a valuable tool to analyze different types of brain disorders and, in the long run, to provide better solutions to mental health issues. We can only reach those goals by means of large-scale analysis, acquiring new MRI datasets, and building strong collaborations between labs around the world. We really believe that by bringing together people and data, we can change the field. Let's start working together, let's connectomize!