Here are our core beliefs that have defined our approach towards breaking barriers in the use of biomedical molecular data.
Biomedical molecular data is available, but not usable.
The growth of biomedical molecular data is exponential at present. This widespread availability of data does not, however, equate to its usability. This is encapsulated by Eroom’s law that states that the process of drug discovery has become slower and more expensive over time, despite abundant data availability.
The use of biomedical molecular data is often limited to a single data type.
High-throughput biomedical molecular data has generated significant medical advances. However, the use of data for insight discovery is often skewed towards one data type. Individual data types cannot provide a holistic snapshot of the biological complexity underlying human diseases. Drug discovery is fundamentally a systems biology problem and requires the integration of multiple types of data.
The implementation of ML algorithms to derive data-driven insights is on the rise.
With the increase in data, there has also been a corresponding surge in publications focused on developing ML algorithms for insight discovery. ML can be used to automate routine, low-level analyses, and help scientists draw deeper insights from the underlying data.
An ML model is as good as its training data. The more relevant the data, the better the accuracy of these models.
Public and proprietary biomedical molecular data is predominantly available in a semi-structured form. This poses a significant challenge to deploying effective and accurate ML models as data quality is of the utmost importance for this process. There is a need for high-quality, ML-ready biomedical molecular data to enable data-driven R&D teams to discover meaningful insights faster.
Data consumption of tomorrow.
Adopting a data-centric approach to research can help scientists formulate new hypotheses by leveraging existing data from publications and controlled repositories, validate pre-conceived hypotheses using relevant previously generated data, and facilitate integrated analysis of diverse data types. Polly empowers multi-disciplinary data-driven drug discovery teams in the biopharmaceutical industry by delivering ML-ready biomedical molecular data.