Tuesday, December 24, 2024

Enhancing Cancer Drug Development: How Big Data and Google BigQuery Detect Bacteria

Developing new drugs is both risky and expensive. Taking a new drug through clinical trials can cost billions, and often, there’s no guarantee of success. Sometimes a drug proves effective in one region but fails in another due to various factors.

One major player in this uncertainty is the bacteria living in our bodies. Everyone has a unique mix of bacteria, and these microbes significantly influence how medications work—or don’t work. This is especially critical in cancer treatments, where bacteria in tumors can block essential therapies.

Understanding this complex relationship is crucial for researchers and healthcare providers. With the staggering cost of developing a new drug, which can reach up to $2.6 billion, accurately modeling how bacteria interact with medications is essential.

BioCorteX is a research company focusing on this challenge. They apply advanced data science to study how bacteria affect drug candidates, particularly in oncology and antibody-drug conjugates. By uncovering how bacteria disrupt treatments, BioCorteX aims to boost the success of drugs in clinical trials, which could shorten development timelines and improve patient outcomes.

Co-founder Nik Sharma expresses the motivation behind the company. He points out that clinicians often notice how differently patients respond to treatments, but understanding the “why” can be frustrating. BioCorteX aims to change the narrative around drug interactions and bacteria’s crucial role in health.

In clinical trials, a drug might succeed in one population but stumble in another, mainly due to differing bacterial compositions among individuals. To explore this, Sharma teamed up with Mo Alomari, a Rolls-Royce engineer. They leveraged Alomari’s expertise in modeling systems to address the vast number of variables involved in studying drug-bacteria interactions.

They decided to model these interactions “in silico,” using computational resources. BioCorteX has constructed one of the largest biological knowledge graphs, showcasing approximately 15 to 16 billion connections between bacteria and drug candidates. This complexity far exceeds the capability of standard databases, prompting BioCorteX to build its own using Google’s BigQuery.

The knowledge graph comprises about three billion nodes and 16 billion edges, custom-designed to handle large data sets. When a scientist needs fresh data, BioCorteX can process it multiple times a day, often in around 20 minutes.

They run data from pharmaceutical companies through this graph to assess potential bacterial interference with drugs, determining how these interactions might affect effectiveness across different patients. Sharma explains that while some drugs are compatible with certain individuals, they could be incompatible for others, and BioCorteX can pinpoint these interactions quickly and efficiently.

This approach not only saves time and money compared to traditional clinical trials but also extends beyond new drugs. BioCorteX also examines previously unsuccessful drugs, known as “out-licensed” assets, to identify any hidden interactions that may have contributed to their failure.

As the global landscape of drug development evolves, BioCorteX’s modeling capabilities become increasingly vital. They can simulate scenarios across different countries, comparing results from various phases of studies in locations like Australia, the U.S., and Europe.

While the current emphasis is on oncology, BioCorteX isn’t limited to cancer research. They are also exploring applications in studying viruses and fungi and have even ventured into consumer health. Their technology provides insights into unavoidable bacterial interactions with drugs. Pharmaceutical companies face a critical choice: either acknowledge this reality or continue down the path with a staggering 96% failure rate in drug development.

Sharma envisions a future where they can ensure the right drug is delivered, first time, for every patient.