"Alone we can do so little, together we can do so much." --Helen Keller
At SVAI, we want to facilitate and support a research community for underserved patients -- focused on the undiagnosed and rare disease community.
More specifically, we're interested in the expandability of collaborative research and new models of how discovery happens. Our research series started with one patient and one genome and 150 engineers and researchers looking for new insights to help push the body of knowledge forward for a rare neuro-cancer: NF2. This gathering showed incredible results and showed promise for how short bursts of research could be useful and integrated into established laboratories. Since then, we've hosted another successful AI Genomics hackathon focused on p1RCC, and looking forward to building a more extensive research pipeline.
Contribute to real, ongoing patient cases.
Create interdisciplinary networks of computer scientists and biologists.
Learn and develop skills in AI/ML, computational biology and genomics.
Build an open community for collaborative biomedicine discovery.
This scientific and technical educational platform aims to improve the transparency, interpretability, and effectiveness of the community research done at SVAI.
We're trying to achieve this goal tackling four major bottlenecks in doing the research:
Understanding the Data
Working on the Cloud
Understanding the biological aspect of the research
Understanding the artificial intelligence aspect of the research
By trying to provide resources to tackle these major bottlenecks, we can then effectively deep dive into illuminating relevant tools and approaches. The hope is then that with the help of your talented mind collaborating with other talented minds in this community, we creatively repurpose and utilize these relevant tools and approaches to tackle any of these three major areas of research (relevant to the patient):
fundamental biological processes underlying human diseases
treating, or developing new treatments.
Below, we'll dive into some summary on current deep learning research efforts. Most of the information is derived from this paper.
A key challenge in biomedicine is the accurate classification of diseases and disease subtypes.
Several studies have used deep learning methods to better categorize breast cancer patients: for instance, denoising autoencoders, an unsupervised approach, can be used to cluster breast cancer patients , and CNNs can help count mitotic divisions, a feature that is highly correlated with disease outcome in histological images . Despite these recent advances, a number of challenges exist in this area of research, most notably the integration of molecular and imaging data with other disparate types of data such as electronic health records (EHRs).
This domain includes modelling biological processes and integrating multiple types of omic data, which could eventually help predict how these processes are disrupted in disease.
Major applications include one-dimensional CNNs and RNNs are well suited for tasks related to DNA- and RNA- binding proteins, epigenomics and RNA splicing. Two- dimensional CNNs are ideal for segmentation, feature extraction and classification in fluorescence microscopy images . Other areas, such as cellular signalling, are biologically important but studied less-frequently to date, with some exceptions .
Although the application of deep learning to patient treatment is just beginning, we expect new methods to recommend patient treatments, predict treatment outcomes and guide the development of new therapies.
One type of effort in this area aims to identify drug targets and interactions or predict drug response. Another uses deep learning on protein structures to predict drug interactions and drug bioactivity . Drug repositioning using deep learning on transcriptomic data is another exciting area of research . Restricted Boltzmann machines (RBMs) can be combined into deep belief networks (DBNs) to predict novel drug–target interactions and formulate drug repositioning hypotheses [38,39]. Finally, deep learning is also prioritizing chemicals in the early stages of drug discovery for new targets .
The following are project backgrounds. For the analyses, results, and ongoing work, click on the project titles!
The main issue with NF2 is that patients develop tumors in the central nervous system, slowly knocking out senses and body functions (hearing for most, facial paralysis and sight/muscles/sensory functions for some). An NF2 patient and serial entrepreneur, Onno Faber, shared his genomic data as well as his brothers to help co-host SVAI's first AI Genomics Hackathon involving hundreds of artificial intelligence engineers and life sciences researchers. The event was additionally hosted by Google, who provided $150,000 worth of processing power and a site for the mass collaboration at Google Launchpad in San Francisco.
After our successful first hackathon event, NF2, a key hackathon contributor, Bill Paseman, reached out to us: he was diagnosed with stage IIIB papillary renal-cell carcinoma type 1 (p1RCC).
Bill’s diagnosed disease falls under the carcinoma family: Papillary renal-cell carcinoma (pRCC). pRCC accounts for between 15 to 20% of all kidney cancers. It occurs in the cells lining the small tubules in the kidney that filter waste from the blood and make urine. Little is known about the genetic basis of sporadic pRCC, and no effective forms of therapy for advanced disease exist.
Realizing Bill was racing against time, SVAI jumped into deep conversations with Bill and after eight months, SVAI announces its second computational cancer genomics event in SVAI’s Collaborative Research Series, p1RCC, in partnership with RareKidneyCancer.org, Salesforce, Google, NIH, NCBI and more.