Cells are bursting with information, from the 3.1 billion base pairs that make up the human genome to the multitudes of biochemical pathways that transform one protein into another. Even the way tissues look under a microscope can hold important clues when tackling diseases like cancer.
However, teasing out the clinically relevant details from a sea of information remains an ongoing mission for cancer researchers. A treatment’s efficacy could be influenced by highly-activated genes in tumours, or by the changing functional states of immune cells. Excitingly, the advent of innovative computational tools is padding the cancer researcher’s arsenal, enabling them to better study the disease across biological scales.
These tools include single-cell bioinformatics, which grants scientists like Mai Chan Lau a more intricate view of the dynamic molecular features, plasticity, development and behaviours of individual cells. Spatial omics technologies take that view a dimension further by shining a spotlight on these interactions in a physical tissue space—effectively framing molecular signals within locational contexts.
Backed by these emerging technologies, Lau concurrently leads her own group as an Assistant Principal Investigator at A*STAR’s Bioinformatics Institute (BII) while heading the Singapore Immunology Network (SIgN)’s Computational Immunology Platform. By harnessing the power of artificial intelligence (AI), Lau and colleagues and her team aim to improve cancer diagnostics and treatment by rummaging through elaborate molecular maps of cancer-immune interactions and translating them into clinically applicable insights.
In this interview with A*STAR Research, Lau reflects on the key people and experiences that shaped her career from trainee to supervisor, and highlights the importance of AI-enabled research in cancer immunology.
Tell us about your scientific journey.
My interest in biomedical research started during my undergraduate final year project at the National University of Singapore with Lakshminarayanan Samavedham. There, I developed neural network models—a form of machine learning—to predict appropriate drug doses for cancer treatment. This project fuelled my passion for AI, and I became convinced that mastering big data was crucial to advance AI research.
Subsequently, I decided to focus on high-performance computing based on graphics processing units (GPUs) for my PhD project under Rajagopalan Srinivasan. That experience solidified my technical skills and belief in the transformative power of AI in handling large datasets. As I was eager to apply my computing expertise to biomedical research, I was fortunate to then join SIgN, where I dove into single-cell bioinformatics in Jinmiao Chen’s lab and received valuable mentorship from Bernett Lee.
During my second postdoctoral position with Shuji Ogino at Brigham and Women’s Hospital in Boston, I realised that it isn’t enough for an immune cell to have the appropriate functional states to target tumour cells. Rather, the physical proximity of these cells is equally crucial; they need to be in reach of their targets. As such, I turned to integrating knowledge from histopathology: the microscopic examination of tissues for disease diagnostics. These combined experiences in single-cell immunology and molecular pathology stimulated my interest in tumour-immune interactions.
How can spatial technologies help tackle cancer?
Simply put, spatial technologies reveal the spatial relationships of cells: how they interact with each other in disease contexts. They fill in key gaps by capturing the significant cell-cell interactions that influence cancer development, progression, drug response and resistance. Such unprecedented insights can help greatly refine biomarkers, enabling more accurate diagnosis and patient stratification. An understanding of cellular and spatial contexts can also create opportunities to identify new treatment targets.
When these advanced cellular and molecular-level insights are integrated with decades of established histopathological knowledge, their advantages are magnified—they can help optimise existing histological or single-molecule biomarkers, which paves the way towards better personalised cancer treatments.
Tell us about your work on the H&E2.0 platform.
While advanced spatial technologies show great promise for cancer immunology research, their high cost and specialised skill requirements significantly hinder their accessibility and potential clinical impact. To address this, I am working with Joe Yeong, my former mentor and a pathologist at the Institute of Molecular and Cell Biology (IMCB), to co-develop the H&E2.0 platform.
H&E2.0 has two aims: the first is to provide interactive, integrated visualisations of molecular signals—obtained from advanced spatial technologies—in haemotoxylin and eosin-stained (H&E) tissue samples at high-resolution views. For reference, H&E is a gold standard in histopathology; the platform’s ability to seamlessly visualise data would be of great help to clinicians, histologists and pathologists that seek to interpret and appreciate new molecular insights of clinical interest.
The platform’s second aim is to improve access to advanced spatial omics data for the research community, especially for those in resource-limited settings. To that end, we’re training H&E-based generative AI models for H&E2.0, using spatial omics data acquired from the same H&E tissue sections as training ground-truth. By synthesising molecular signals from H&E tissue space, our platform also enables a seamless integration of data at multiple biological scales.
We think H&E2.0’s visualisation capabilities provide essential interpretations that can bring AI-predicted biomarkers into clinical settings. In future, we intend to make AI-capable H&E2.0 a publicly available web tool through our collaboration with Minh Nguyen and Chandra Verma from BII.
How did your research journey influence your leadership style?
My diverse experiences in different labs taught me the importance of focusing on the career growth of my team members and understanding their interests, whether they are oriented towards academia or industry. I strive to give them chances to develop leadership skills by allowing each of them to lead projects that align with their interests, such as bioinformatics tools, production pipelines or AI projects.
Recognising the importance of showcasing our work, I ensure my group has ample opportunities to present their research at various venues, and to give them credit both verbally and in authorship. Most importantly, I aim to cultivate an open and receptive environment where every team member feels safe to provide feedback and share ideas.
How has A*STAR supported you as an early-career researcher?
As I have just started my own lab, A*STAR’s support has been invaluable in providing crucial access to various resources. These include computing facilities and mentorship from senior BII and SIgN group leaders, particularly BII Executive Director, Sebastian Maurer-Stroh; and SIgN Executive Director, Kong-Peng Lam. The agency has also facilitated interactions with management; at a Group Engagement Session, I had the chance to discuss grant mechanisms with Huck Hui Ng, the Assistant Chief Executive of Research and Talent Development.
In turn, mentoring students through A*STAR-funded internships has enabled me to expand my research more effectively and enhance my leadership skills. In addition, platforms like the A*STAR Research publication and Science Brew—organised by Andy Hor, Deputy Chief Executive (Research), and his team—have given me opportunities to publicise my work. These experiences have been important in my growth as a researcher and a leader.
What advice can you share with aspiring scientists?
My key motivation for staying in research—for nearly 10 years, to date—is a genuine passion for my work and a strong belief in its potential societal impact. I think that one crucial trait for a successful researcher is the ability to recover from failures and rejections. It is important to recognise that it takes time to learn and grow through feedback from others. Embracing this learning process is essential for long-term success in research.
