So many people are dying, it is devastating! There is no time to make a vaccine or start the development of a new drug. The research machine is slow, but for COVID, we have rapid data being released and with so many patients, many drugs have already been used in patients either as a test or due to the fact that patients were already on other drugs when they became ill for other reasons. Many approved drugs have alternative functions (side effects/ off target effects). A large number of approved drugs have demonstrated in vitro effects inhibiting COVID19, with more predicted in silico to have effects. Unfortunately, the drugs currently being tested are not working well in the COVID patients. The answers are in the data and therefore, we wanted to harness this information to identify the best drugs to progress to clinical testing.
There is a lot of data on approved drugs that we have worked with to identify those that have a better chance of working it is a balance between dosing, side effects and the ability to kill the virus.
What it does
It provides an interface filtering drug options based on effectivity in vitro and safe dosing information to help identify the best drugs for testing in patients. Another layer is collecting data from COVID survivors to see if the drugs they were taking before being infected or drugs that were prescribed are associated with less severe disease. By analysing data from survivors, we may find a drug that has been overlooked and therefore this could be the key to solving the COVID crisis. These two pieces of information will let us identify the best treatment options quickly.
What we did this weekend
- Built a website front-end to provide clinicians and biomedical researchers with a heavily curated short-list of key drugs used in the treatment of COVID-19.
- Analyzed scientific literature to extract key metrics for clinical treatment decisions (Cmax, IC50).
- Developed an analysis tool that identifies the therapeutic window, graphically representing this for easy identification of useful drugs.
- Developed a survey form for global use to allow direct community engagement in the collection of data.
- Developed data analytics tools that identify new drugs used by survivors both before and during COVID treatment, graphically representing these also.
How I built it
We built it using the processing and curation of scientific literature data sources to provide the key metrics for drug leads to be identified. We used Rshiny, Python, GCP, ASP.Net and HTML5 to make the website and interactive tables and plots. We used R Machine platform for sanity check for All data captured via user response. Drug Related data captured from various authentic sources and sanitized via M L Sanity check Model. Integration of User Response data from different channel to One common Repository. Create R shiny App and deployed on shinyapps.io The client facing website is created with html5 /Asp.net 3 D charts were introduced.
Challenges I ran into
Difficulty finding reliable sources for pharmacokinetics data and intense time pressure.
Accomplishments that I'm proud of
Designing a platform in such a short period of time with a new globally connected team.
What I learned
Connecting the data is really important, but also visualization is imperative to allow researchers and doctors to access the data in a meaningful way.
What's next for Drugs4COVID-finding the needles in the haystack
Continue to collect data as it emerges in the literature. Very little clinical trial data has been released, however, this data is starting to be released, so we will integrate this data into the platform. In order to continue the project, we would need some funding to host the site and manage it as the data grows. This could be achieved through small grant funding. Further, we will develop visualization tools in response to user experience feedback. As the data volume and variety increases, more sophisticated AI and machine learning algorithms will need to be developed to gain the most benefit from our data. Furthermore, we would like to link it to other databases such as ChEMBL and PubChem. To add in the extra functionality, developers would be required. In the long term when COVID has passed, we would be able to adapt the platform for other diseases that desperately need drugs.