We've updated our Privacy Policy to make it clearer how we use your personal data.

We use cookies to provide you with a better experience. You can read our Cookie Policy here.

Advertisement
Leveraging AI To Identify and Target Novel Cell–Cell Interactions
Industry Insight

Leveraging AI To Identify and Target Novel Cell–Cell Interactions

Leveraging AI To Identify and Target Novel Cell–Cell Interactions
Industry Insight

Leveraging AI To Identify and Target Novel Cell–Cell Interactions

Credit: Phenomic AI

Want a FREE PDF version of This Industry Insight?

Complete the form below and we will email you a PDF version of "Leveraging AI To Identify and Target Novel Cell–Cell Interactions"

First Name*
Last Name*
Email Address*
Country*
Company Type*
Job Function*
Would you like to receive further email communication from Technology Networks?

Technology Networks Ltd. needs the contact information you provide to us to contact you about our products and services. You may unsubscribe from these communications at any time. For information on how to unsubscribe, as well as our privacy practices and commitment to protecting your privacy, check out our Privacy Policy

Technology Networks recently had the pleasure of speaking with Sam Cooper, CEO and co-founder of Phenomic AI. Cooper discusses how the company is leveraging advanced machine learning techniques to interrogate cell–cell interactions responsible for driving many disease mechanisms.

Of particular interest is the tumor stroma, which has been shown to be associated with resistance to systemic therapies. The stroma
comprises the basement membrane, fibroblasts, extracellular matrix and vasculature, as well as various cell types including cancer-associated fibroblasts (CAFs). CAFs have been shown to facilitate cancer progression and therefore warrant further interrogation – Cooper elaborates on Phenomic AI’s work in this area.

Laura Lansdowne (LL): Can you tell our readers a little about yourself, your career and how you came to found Phenomic AI in 2017? 

Sam Cooper (SC):
I completed my PhD split between Imperial College London and the Institute of Cancer Research, (UK) under the supervision of Prof. Robert Glen an early pioneer in chemo-informatics, and Prof. Chris Bakal, a leader in high-content screening. Being part of two great labs, both at the forefront of using machine learning (ML) to solve problems in biology and drug discovery, put me in a good position to understand where ML could be applied to maximum effect in pharma pipelines.

Toward the end of my PhD I got involved with the CytoData society, a small community of high content image analysts nucleated by Anne Carpenter’s group at the Broad; known for the Cell Painting Assay. I met my co-founder Oren there who was applying deep-learning (DL) technologies to high content analysis, and that solved a number of issues I’d had with traditional ML approaches over my PhD. He was starting a company up in Toronto – Phenomic AI, so I sorted out a visa for Canada, jumped over the pond and Phenomic began in Oren’s living room.

LL: Cell–cell interactions are known to drive many disease mechanisms but are often neglected in drug discovery. Why is this? 

SC:
Interactions between different cell types are often neglected as most high-throughput drug discovery workflows focus on cells grown in isolation, or with the addition of exogenous agents, such as, cytokines. This means cell–cell interactions have typically emerged from lower throughput e.g., mouse knockout studies. There thus lies a significant opportunity to use more brute force/high-throughput approaches to identifying disease driving cell–cell interactions through the use of complex multicellular in vitro models. 

However, major blocks exist, preventing widespread uptake of multicellular in vitro models. Specifically, deconvoluting the effects of target inhibition (when many cell types are involved) can be very difficult with traditional endpoints. For example, in the absence of single-cell readouts many counter screens need to be run to determine which effects are on-target vs off-target. With single-cell endpoints, such as high-content imaging or high-throughput flow cytometry, deconvolution is simple – the challenge is then analyzing large high-dimensional datasets effectively. This is where ML/DL comes into its own, and where we think a significant opportunity to identify novel biology lies.

LL: Can you elaborate on the significance of tumor stroma in relation to immunotherapy response? 

SC:
The tumor stroma is increasingly being recognized as a key driver of resistance to immune checkpoint inhibitors. A lot of this is being driven by our increasing understanding of transforming growth factor beta (TGF-β) and the role it plays in driving the formation of a suppressive stromal wall that locks in our immune-cells and prevents them from killing cancer cells. The importance of this was demonstrated in recent work by Shannon Turley’s group at Genentech where they show that whilst TGF-β treatment alone had little effect on tumor growth in vitro, TGF-β significantly enhanced the effects of immune checkpoint inhibition (PD-L1 blockade). A number of clinical trials investigating the combination of TGF-β therapy with immune checkpoint inhibitors are now ongoing. 

We think that TGF-β is just the tip of the iceberg, and that a number of stroma factors are driving immunotherapy resistance and that the relative importance of these varies between different cancers. With precise biology, we think stroma biology will lead an inevitable second wave of immune therapies, that this time will open up solid tumors to attack.

LL: How is Phenomic AI using artificial intelligence and machine learning approaches to develop stroma-targeted medicines? 

SC:
We use ML to process complex multi-cell experimental data emerging from our internal discovery efforts, and those of our academic collaborators. Unlike many biotech's, our DL tools let us process highly complex imaging data from co-cultures of fibroblasts and cancer cells and immune cells. This system is ideally suited to interrogating cancer cell-secreted targets that induce cancer-associated fibroblasts (CAFs) or are being secreted by CAFs and in turn are suppressing other immune cell types.

Moreover, we’ve recently doubled down on using in vitro scRNA endpoints to further understand the mechanistic changes to cell-types that our target inhibitors are causing. Beyond this we’ve now shown that DL tools can be used to map in vitro data directly back to human tissue datasets, giving us key reference points for understanding: (1) How relevant our assays are; and (2) Whether are drugs are pushing cells toward more desirable inflammatory states.

Overall, we think these tools will be key to solving the long-standing puzzle that has been Stromal/CAF biology.

LL: Can you tell us more about the two cancer drug targets discovered using the company’s platform?

SC:
Having joined forces with renowned drug hunter, Mike Briskin, earlier this year, we’re now pursuing a set of stromal targets, that we believe are playing key roles in supporting the tumor microenvironment in distinct cancers.

Our lead target looks to be involved in both transforming fibroblasts into CAFs and suppressing immune cells, in a similar vein to TGF-β. We’re now running in vivo proof of concept studies to further establish the role we think it’s playing in driving resistance to immune checkpoint inhibitors; results are eagerly anticipated this quarter.

With a number of other targets lined up, we expect this year is going to be transformative for the company.

Sam Cooper was speaking with Laura Elizabeth Lansdowne, Managing Editor for Technology Networks. 

Meet The Author
Laura Elizabeth Lansdowne
Laura Elizabeth Lansdowne
Managing Editor
Advertisement