new to this stuff.
i’m in IT (operations) and have a friend with Ulcerative Colitis.
They’ve started tracking and i’m looking for implementations of an analysis system for this…
Has anybody seen anything available?
As i said i’m in IT so could easily put something together, but i dont really know what it is i should be analyzing
I’m reading through a bunch of papers and trying to build up an idea of what’s out there. though this stuff is super technical.
I’ve seen larry smarr’s talk from 2012 which is awesome for his analysis of so many different variables.
I assume by now we could run his supercomputer stuff in the azure cloud for peanuts.
Larry’s approach is awesome and also pretty daunting. I’ve seen the resources he has down at Calit2 for genomic analysis and it’s far outside the realm of ordinary mortals. Yes, you could probably mobilize the computing power, but the gating factor is actually the expertise required to develop meaningful hypotheses and design good experiments. If you do go this route, there are some folks here (hi @sprague) who have gotten deeply into their microbiome data.
Another approach would be to work with your friend to focus on one or two symptoms that are discrete enough and variable enough to form the basis of an ongoing self-research project based around self-tracking these symptoms. The first step, after identifying the phenomenon, would be the test an observation protocol: is the phenomenon trackable and variable enough to have some kind of interesting signal. The second step would be an observation phase: Track the phenomenon for some time and then, reflecting on these observations, develop some ideas about what’s influencing the colitis.
Starting with these observations can really clarify the research path.
thanks Agaricus i really appreciate your thoughtful response!!
I’ll definitely take your suggestion to work on a couple of symptoms with proper scientific method as it’s good (and accessible!) science.
the developer/hacker-side of me is really attracted to the idea of seeing what kind of group effort could be done on the compute side. e.g. what if anybody could just grab a recipe (kubernetes helm chart) to spin up their own amazon/azure compute cluster and run their own data against the e.g. 1000IBD dataset. convince some bona-fide scientists to provide some algorithms/methodology that can be packaged up and run on this cluster and just play “data miner”.
@sprague definitely very interested in any details regarding your analysis
