By Ben Samudio
Developing medicine is a time-consuming, expensive, and error-prone endeavor [1] that is typically conducted by private research groups whose work is not usually shared with the public. As a result, many diseases—especially rare and infectious ones—lack available medicines for treatment [2]. At the same time, previously unencountered and reemerging infectious diseases are increasing in diversity and prevalence at an alarming rate across the globe [3]. For some reemerging diseases such as tuberculosis, medicines do exist. Yet these medications are rapidly becoming ineffective as disease-causing microbes continue to adapt and evade our defenses [4]. Newly-encountered diseases can jump across species to humans and cause devastating, untreatable outbreaks. For example, severe acute respiratory syndrome (SARS)—caused by a newly-discovered coronavirus (CoV)—led to a global outbreak in 2003 that reached over a dozen countries, infected about 8,000 people, and killed nearly 800 individuals in approximately an eight-month span [5]. While no new cases of SARS have been reported since 2004, the malady can surface again and severely threaten human health. No SARS-specific therapeutics currently exist to treat these disease [1]. We must develop medicine quicker and more efficiently if we hope to meet the mounting challenge posed by these types of infections.
Open-access Crowdsourced Medicine Discovery
Crowdsourcing has shown tremendous potential for addressing some of the most pressing challenges of our time [6]. Through collective efforts, it promises to produce innovative solutions at an accelerated pace. StudentsGiveHope.org is one of the first open-access crowdsourced efforts to power medicine discovery. We involve students and citizen scientists to help find treatments that otherwise might not have been pursued. This process is analogous to selecting just the right key among many to open a lock. If one explicitly knows the dynamics of the lock’s interior, it might be possible to reverse-engineer the key’s design. In this analogy, the lock is a disease target, the key is a small molecule, and opening the lock is akin to successfully treating the disease. In SARS-CoV, for example, a protein known as 3CL protease (3CLpro) is crucial for the virus to replicate and the disease to progress. “Plugging up” this protease with a medicine—or small molecule—could stop SARS and allow the host time to clear the infection.
Our Current Approach
We start by developing both computational and physical models of a disease target. The computational model simulates how the disease target interacts with a small molecule — like modeling how the pins in a lock accommodate the patterns on a key. Those small molecules interact favorably with the disease target, and we evaluate them more closely for their potential to become actual medicine. The physical model is a scaled-up representation of the disease target that is made of papier-mâché and colored to highlight important features (see Figure 1). Participants create ideas for medicine by piecing together plastic molecule sets and using the physical model as a guide. They then upload their ideas to the computational models, which are hosted online. These models return feedback via a “molecule report card,” “therapeutic grade point average (tGPA),” and augmented reality visuals. The goal is to achieve the highest tGPA possible. Participants continuously refine their ideas based on the feedback they receive, and all of their work is made freely available to inspire medicine development and collaboration. Through this process, participants learn science, technology, engineering, and mathematics; become inspired by applying what they learn towards helping others; and gain important research skills. Patients and researchers benefit because advancements towards treatments are openly shared.
Figure 1. Segmentation of DNA as an example of the first step in constructing a physical model. A side view of DNA’s double helix is displayed on the left. The elements that compose DNA are colored as follows: red signifies oxygen, dark blue is nitrogen, light blue is carbon, tan is phosphorous, and white is hydrogen. DNA is sliced into evenly-spaced segments from top to bottom along its long axis. These segments are visible on the right, looking down along the long axis, and colored as follows: violet signifies the hydrogen bond acceptor, orange is the hydrogen bond donor, and gray is everything else. We then scale the segments and use them to make a mold from which the papier-mâché physical model of DNA is cast.
Future Direction and a Call to Action
Through StudentsGiveHope.org, participants have designed thousands of ideas for medicine to treat diseases like SARS, antibiotic resistant bacterial infections, and cancer. The methodology that underlies the computational evaluation of these ideas is molecular mechanics (MM), which can be faster but less accurate than quantum mechanics (QM). We ideally want an approach that is as fast as MM and as accurate as QM. Such an approach would provide clues to optimize our search, as well as greatly enhance our ability to quickly and accurately test ideas for medicine. We have partnered with Silvia Crivelli and Rafael Zamora-Resendiz, both of Lawrence Berkeley National Laboratory, to reach this goal. We are developing a platform called OpeNNdd—open neural networks for drug discovery—that will eventually replace MM as the method to evaluate ideas for medicine. As a proof-of-concept, we have produced a graph convolutional neural network (GCNN) model of 3CLpro and demonstrated that it can test ideas for medicine as quickly as MM. Our next step is to train this model on QM data where we anticipate that the GCNN will remain as fast as MM but approach the accuracy of QM.
Open crowdsourced medicine discovery depends on volunteers who work together and search for medicines. If you would like to become involved, please visit StudentsGiveHope.org and inquire through our contact page.
The author presented this research during a minisymposium at the 2019 SIAM Conference on Computational Science and Engineering, which took place earlier this year in Spokane, Wash.
References
[1] Everts, M., Cihlar, T., Bostwick, J.R., & Whitley, R.J. (2017). Accelerating Drug Development: Antiviral Therapies for Emerging Viruses as a Model. Ann. Rev. Pharmacol. Toxicol., 57(1), 155-169.
[2] TEDMED 2012: We need better drugs – now. TED Talks. Retrieved from https://www.ted.com/talks/francis_collins_we_need_better_drugs_now
[3] Smith, K.F.. Goldberg, M., Rosenthal, S., Carlson, L., Chen, J., Chen, C., & Ramachandran, S. (2014). Global Rise in Human Infectious Disease Outbreaks. J. R. Soc. Inter., 11, 20140950.
[4] Sohail, M. (2006). Tuberculosis: A re-emerging enemy. J. Mol. Genet. Med., 2(1), 87-88.
[5] Centers for Disease Control and Prevention. (2017). SARS Basic Fact Sheet. Retrieved from https://www.cdc.gov/sars/about/fs-sars.html
[6] Khare, R., Good, B.M., Leaman, R., Su, A.I., & Lu, Z. (2016). Crowdsourcing in biomedicine: challenges and opportunities. Brief. Bioinform., 17(1), 23-32.
 |
Ben Samudio is the director of the Hope Shared Foundation (StudentsGiveHope.org) and an adjunct assistant professor of chemistry at Sierra College in Rocklin, California. He is passionate about enriching education and providing hope to patients by involving students and citizen scientists in open-access medicine discovery. |