Immunai CTO and Co-Founder Luis Voloch on Using Deep Learning to Develop New Drugs - Ep. 176

Data Quality

The importance of data quality in biological research cannot be overstated, especially when dealing with the complexities of the immune system. Luis Voloch emphasizes that generating high-quality single-cell genomics data is crucial for effective machine learning applications in biology. He notes that the data is often noisy, requiring close collaboration between biologists and machine learning experts to refine lab protocols and improve data quality.

One of the main decisions that we made early on in the company was that we wanted to become the world leaders, not just in analytic this kind of data, but actually in generating the kind of data that we focus on.

--- Luis Voloch

While synthetic data holds promise for the future, Voloch points out that more understanding and high-quality biological data are needed before it can be reliably used 1 2.

   

Deep Learning

Developing deep learning frameworks tailored to biological data presents unique challenges and opportunities. Luis Voloch discusses the task of annotating cells into types and states, a process that involves handling vast amounts of complex data. This data, derived from millions of cells, requires sophisticated algorithms to classify cell types automatically, reducing human bias and increasing scale.

One of the challenges actually that we have is that there aren't many deep learning frameworks which are made for the kind of data that is just like that.

--- Luis Voloch

Voloch highlights the need for advancements in both data generation and computational power to efficiently analyze this data, underscoring the pioneering work being done to develop suitable deep learning frameworks 3.