41
$\begingroup$

Is anyone aware of a medicine/vaccine related challenges with a good problem definition and some available input datasets, ideally related to COVID-19? What kind of constraint solving challenges would that be? Protein folding?

If there are any such constraint solving problem, we could drop them in GitHub and maybe crowd solve them in the community. If public life gets as limited as in China and Northern Italy, we'll have plenty of time on our hands anyway :)

UPDATES

The primary goal right now is to find a solid problem definition for COVID-19 protein folding. The most promising references so far:

  • FoldIt - not investigated yet (credits: Mast)
  • Folding@Home - asked on forum, not yet investigated further (credits: Sebastian)
  • GISAID datasets - not investigated yet (credits: Joe)

If you're investigating any of these to find a problem definition, add a comment below. Also edit this question with your results.

Other use cases

| improve this question | |
$\endgroup$

10 Answers 10

Active Oldest Votes
16
$\begingroup$

Actually, yes, and Folding@Home is doing this right now.

https://foldingathome.org/2020/02/27/foldinghome-takes-up-the-fight-against-covid-19-2019-ncov/

We need your help! Folding@home is joining researchers around the world working to better understand the 2019 Coronavirus (2019-nCoV) to accelerate the open science effort to develop new life-saving therapies. By downloading Folding@Home, you can donate your unused computational resources to the Folding@home Consortium, where researchers working to advance our understanding of the structures of potential drug targets for 2019-nCoV that could aid in the design of new therapies. The data you help us generate will be quickly and openly disseminated as part of an open science collaboration of multiple laboratories around the world, giving researchers new tools that may unlock new opportunities for developing lifesaving drugs.

And another bit of info from the link:

One protein from 2019-nCoV, a protease encoded by the viral RNA, has already been crystallized. Although the 2019-nCoV spike protein of interest has not yet been resolved bound to ACE2, our objective is to use the homologous structure of the SARS-CoV spike protein to identify therapeutic antibody targets.

| improve this answer | |
$\endgroup$
10
$\begingroup$

There is the following data set on the number of infections in China provided by Isaac Lin from the University of Hong Kong. https://github.com/BlankerL/DXY-COVID-19-Data/blob/master/README.en.md

I am not sure how reliable the data is, but it’s the best I could find. And from a first look, the numbers are close to what the German news are communicating. Isaac has created a web crawler which is also available on GitHub https://github.com/BlankerL/DXY-COVID-19-Crawler/blob/master/README.en.md

I was made aware of this data set when looking for visualizations of time trends and model predictions of confirmed cases on his website. E.g., https://ncov2020.org/en/en/predict-2020-03-05/

| improve this answer | |
$\endgroup$
  • $\begingroup$ This doesn't appear to answer the question. The question asked for an optimization challenge with a dataset. I don't see any optimization challenge articulated here. $\endgroup$ – D.W. Mar 7 at 5:36
  • 1
    $\begingroup$ I agree, it does not fully answer the question. However, in my experience many people in the field struggle to access realistic data to evaluate their great algorithmic ideas. So, I thought mentioning this data set could still add value to the discussion, although it will not be the final answer to the question. $\endgroup$ – Doegstra Mar 7 at 7:40
10
$\begingroup$

I was thinking of ( no data but):

  • determining which airline routes to block that will minimize the economic consequences given a very low risk of getting to the virus (like a chance constraint or robustness), kind of a stochastic edge removal problem.

  • determining whom and when to vaccine first (whenever they find it). Kind of a stochastic node blocking problem with some time index to minimize the expected number of infected people.

Recently i saw these links that contain some data

| improve this answer | |
$\endgroup$
  • 1
    $\begingroup$ This doesn't appear to answer the question. The question asked for an optimization challenge with a dataset. Your proposed challenges don't come with a dataset, so it doesn't meet the requirements stated in the question. $\endgroup$ – D.W. Mar 7 at 5:37
8
$\begingroup$

We (operations research community in Italy) have just offered our availability to contribute as we can. On our web site https://webgol.dinfo.unifi.it we just starting collecting ideas on how to help. We think there is much that OR can do: assignment (we have strong imbalance between available beds and patients in acute care), transportation, ambulance scheduling, triage optimization, inventory management,....

Any help or suggestion from the community is more than welcome. My group also worked in the past on folding and molecular design, but I think in this moment our emergency is on the logistic side, there is no time, I guess, for a molecule. I mean, a vaccine would be very welcome, but I don't think we can wait, it's simply too late. Anyway, as we really do not know how long all of this will be, we just made available our computing resources and joined the folding@home project https://foldingathome.org/ who is now devoted to finding potential drugs against the virus

Italy has taken impressive and very serious decisions to try to limit the spread; we hope other countries will follow the same path, the earlier the better for all.

| improve this answer | |
$\endgroup$
  • 1
    $\begingroup$ That's great! If there are any problem definitions or datasets available and you can share them, let's crowd solve them with the entire OR community. $\endgroup$ – Geoffrey De Smet Mar 12 at 7:43
7
$\begingroup$

Another optimization related problem (no data) is raised in the following recent article: How America Can Avoid Italy’s Ventilator Crisis.

The formulation of this ventilator distribution problem is the following:

Once industry rapidly scales ventilator production, where should these ventilators be delivered? Which hospitals need them most? How can we build a nimble logistics operation that can rapidly deploy these machines the moment that a shortage appears imminent?

| improve this answer | |
$\endgroup$
5
$\begingroup$

I'm not defining an optimization problem challenge, but providing a dataset I came across when visiting our world in data website. There is this data repository by Johns Hopkins Center for Systems Science and Engineering for COVID-19 cases. The data sources there are from World Health Organization (WHO) and the CDC of China, Hong Kong, US, Australia, Italy, Europe, and more.

Edit:

Amazon is making a public AWS COVID-19 data lake available.

| improve this answer | |
$\endgroup$
4
$\begingroup$

I cannot write regular comments, but the BOINC platform and their Rosetta team are fighting too:

https://www.ipd.uw.edu/2020/02/rosettas-role-in-fighting-coronavirus/

and

https://boinc.bakerlab.org/

The consortium consists of the University of Washington, Howard Hughes Medical Institute and Institute for Protein Design.

Feel free to add to the OP so others can help if they already use BOINC.

| improve this answer | |
$\endgroup$
4
$\begingroup$

Data about the COVID-19 cases in Italy is available on Github from the government:

https://github.com/pcm-dpc/COVID-19

| improve this answer | |
$\endgroup$
3
$\begingroup$

My research group at MIT has collated a dataset of 130+ papers on COVID-19 clinical outcomes, where each paper contains aggregate demographic information/comorbidities/lab data/treatment/clinical outcomes on a group of patients http://www.covidanalytics.io/dataset.

Since we ultimately want to make decisions to combat the spread of COVID, we're currently in the process of using this data and an epidemiological model predicting COVID's spread to define and solve some optimization problems; I'll update this answer with some more details once we finalize these models and make them public.

We'd love to hear if anyone has suggestions for other optimization problems we can solve with this data.

| improve this answer | |
$\endgroup$
1
$\begingroup$

We've recently been working on hospital staff rostering with covid-19 specific constraints, such as these:

enter image description here

There's a working demo on this branch on github. It's open source in java.

| improve this answer | |
$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.