RESOURCES AND PROJECTS

The goal of the Cellular Human Bodymap (CellHuB) project is to understand the functional states of the human body at cellular resolution, in order to advance basic biology, reveal disease mechanisms, detect biomarkers, identify drug targets and improve human health. As a first step in this direction, we are building reference databases of single cell gene expression profiles from multiple normal tissues. In parallel, we are compiling matched sets of single cell profiles from disease-affected tissues. These will be complemented with other single cell readouts such as chromatin openness and DNA methylation, using emerging technologies for multimodal single-cell profiling. We are also developing advanced machine learning algorithms and software tools for clustering cells by functional similarity. These clusters will be used to define human cell types and states de novo, together with their marker genes.

Platinum Asian genomes as a refence for Clinical Genomics in Singapore

The current reference assembly of the human genome was generated using capillary sequencing with read lengths mostly below 1,000 basepairs and biased genome coverage. As a result, the reference genome has numerous gaps. In particular, regions with atypical nucleotide content or highly repetitive DNA are often either missing or incorrectly assembled. This is a major challenge for personalized medicine, because: 1) such regions have been implicated in multiple genetic disorders and 2) they are highly variable across ethnic groups. Moreover, the reference assembly differs significantly from Asian genomes because it was derived from a Caucasian sample. The Platinum Asian Genomes project will use a third-generation sequencing of ultra-long reads (Pacific Biosystems RS II) to assemble highly accurate and complete reference genomes for multiple Asian populations. These "Platinum Genomes" will provide, for the first time, an accurate and reliable reference for genomics research and precision medicine in Singapore.

Singapore 10,000 Genomes Project.

Genetic variation plays an important role in a variety of human diseases and quantitative traits. Due to different genetic makeups and environmental exposures, many genetic findings have shown population-specific characteristics, highlighting the importance of population diversity in human genetic studies. The Singapore population consists of three major ethnic groups, Chinese, Malay, and Indian, which together represent >80% of the Asian population. To empower human genetic studies in Asian populations, the SG10K project will perform whole-genome sequencing (WGS) of 10,000 Singaporeans to 12-13x coverage. Coupling with powerful bioinformatics tools, our study design will enable high-quality genotype calling for a full frequency spectrum of genetic variants segregated in the population. Our main objectives are to (1) comprehensively characterize genetic variation in Singapore population; (2) create a WGS reference panel for accurate genotype imputation in Singapore population; and (3) generate a large control dataset for WGS-based genetic association study of diseases. Upon completion, this study will provide valuable genetic information to facilitate clinical and pharmaceutical research in Singapore populations and will empower genetic studies of Singapore and Asian-centric diseases.

This project is a national effort across multiple institutions. Our collaboration partners include SingHealth Duke-NUS Institute of Precision Medicine (PRISM), Singapore Eye Research Institute (SERI), Centre for Personalised and Precision Health (CPPH), National University Health System (NUHS) and several Translational and Clinical Research Flagship Programmes (cardiovascular disease, Parkinson disease, etc.).

Resistance and Outbreak Tracking in Singapore

The spread of drug resistant bacteria and the emergence of novel pathogens are significant public health challenges worldwide. As a major hub for international trade, business and travel, Singapore is at the crossroads of these international phenomena. By combining cutting-edge genomic technologies with state-of-the-art bioinformatics pipelines, and in partnership with government agencies and hospitals, we are building a sequence database of genes and pathogens isolated in Singapore and its neighborhood. Applications of this database include detection and monitoring of emerging infectious diseases in Singapore (such as the recent Hepatitis C cluster and Group B Streptococcus outbreak); tracking down sources of hospital acquired infections; understanding the dynamics of rising antibiotic resistance; evaluating the impact of the microbiome on health, wellness and disease, and understanding transmission of viral illnesses such as Influenza, Dengue, Norovirus and Adenovirus. The team is also developing analytical approaches to enable definitive molecular epidemiology, transmission tracking, outbreak detection and monitoring, and prediction of antibiotic resistance through integration of genomics datasets from the GERMS platform at GIS.

Expertise in clinical-grade genomics data and software development.

GIS has a robust clinical-grade infrastructure where sequencing data is monitored at the highest levels of laboratory medicine rigor. Our bioinformatics and sample tracking pipelines are subject to quality control and various data checking. Different steps of the sequencing workflow are tracked and open to regular audit (both internal and external). POLARIS has developed a College of American Pathologists (CAP)-grade genomic pipeline to deliver clinical-grade reports to patients in Singapore.

Singapore Oncology Data Portal (OncoSG)

The field of cancer research is advancing at a rapid pace due to the declining cost of next-generation DNA sequencing. Massive DNA sequence data, complex molecular profiles, and rich clinical data are being generated from thousands of cancer patients worldwide. Several such projects are also underway in Singapore. However, considerable computational expertise and resources are required to integrate and mine these datasets. This presents a significant barrier for clinicians and use of this data in cancer precision medicine. To meet this demand, we are developing the Singapore Oncology Data Portal (OncoSG). OncoSG will enable integration, visualization, analysis, and sharing of cancer genomics datasets in Singapore. This will speed-up translation of cancer genomics data into new biological insights and clinical applications.

TOXicity PREdiction for Clinical Innovations in Singapore

Differential drug effects and toxicity will be modelled for SNP variations in albumin (for drug distribution), cytochromes (for drug metabolism), pumps and transporters (for drug influx or efflux), and adverse drug reactions with off-target proteins (for drug toxicity or side effects). BII’s strutucal, chemogenomic, metabolomic, and imaging -based cellular phenotypic profiling platforms will together be brought to stratify medicines and patients towards a distinct predictive precision medicine (P2M) platform. BII and I2R have developed capabilities of transferring the platforms to mobile apps.

PIs: Hao Fan, Igor Berezovsky, Peter Bond, Chandra Verma, Lit-Hsin Loo, Lee Hwee Kuan, Samuel Gan, Ng Siew Bee, Prakash Arumugam, Sebastian Maurer-Strauh, Yoganathan S/o Kanagasundaram

Intelligent software for histology image analysis and biophysical modelling

Diagnosis and prognosis of cancers are major issues for a trained pathologist. Inter-observer variability and tediousness of tissue reading hamper the accuracy of assessment by the pathologist. Clinical outcome can be improved by using data fusion techniques to combine histological slides, gene expression data and clinical parameters. Machine learning methodologies to be use to mine the associations between histological images, gene expression, clinical parameters and disease outcomes. Diagnosis and prognosis of cancers carried out by pathologists will be complemented with an intelligent software to improve accuracy and reduce labour for tissue reading

PIs: Lee Hwee Kuan, Cheng Li, Chiam Keng Hwee, Tan Min Han (collarborator from SGH & IBN)

A*STAR Data Analytics Exchange Platform for federated data

A*DAX is a unified federated data exchange and analytics platform designed to break down existing data silos and unravel invaluable insights from the datasets through advanced data analytics. A*DAX links multiple data sources from heterogeneous data systems, regardless of their locations, or whether they are structured or unstructured, into a unified and well-orchestrated system so that data can be shared with proper control, and analytics algorithms can be applied on the datasets effectively. A*DAX emphasizes on transparency, heterogeneity, and openness.

The current explosions in the volumes of omics data and electronic health records will open up new frontiers for biomedicine. Such opportunities can be captured and progress expedited with an effective data platform to exchange, collaborate, analyze, and visualize the heterogeneous datasets from multiple data owners and users, using the most advanced data analytics and machine learning algorithms -- all of which A*DAX is designed for. A*DAX has already been deployed and tested for Smart City applications. It is now ready for enabling Big Data for Precision Medicine.

Powering data-driven personalized healthcare for everyone

The Data Analytics Department (DAD) at I2R houses a large team of talented data scientists who routinely design and develop novel machine learning techniques to analyze big data for real-world applications. With a unique history of starting out as one of the early Bioinformatics groups in Singapore more than a decade ago, the department excels in cross-disciplinary applied data mining research across multiple industry verticals such as Telecom, Finance, Aerospace, Manufacturing, Energy and Utilities, and E-commerce.

Using advanced machine learning techniques developed in-house, the DAD data scientists are able to uncover hidden insights residing within the large volume of biomedical (omics-level data e.g. genomics, proteomics) and healthcare (results from diagnostics tests, patient vital signs information and unstructured data from electronic medical health records) data. Such insights obtained are integral to the identification of actionable areas which could be targeted to enhance public health in the era of precision medicine. For example, the DAD data scientists have successfully developed machine learning techniques in the study of electronic medical records to more accurately predict patient readmission and to analyze large amount of transcriptomic data to identify the most effective drug/drug combinations for individual cancer patients.