Complex Cellular Phenotype Analysis Group

Drs. Lit-Hsin Loo and Daniele Zink received the Lush Prize 2016 (Science Award)
for their high-throughput in vitro nephrotoxicity models.

Our research is focused on the development of in vitro and computational models for predicting the toxicity and/or targets of chemical compounds. We develop novel cellular phenotypic profiling methods and tools that can automatically classify the effects of compounds with diverse or even unknown chemical structures. These technologies allow us to build highly accurate and scalable in vitro cell-based models for predicting organ-specific toxicities. Our models could be used as high-throughput alternatives to animal testing.




Our current projects include:


Toxicity prediction based on phenotypic profiles: We have developed an approach to predict xenobiotic toxicity using high-throughput imaging of cultured human cells, quantitative phenotypic profiling, and machine learning models. We have applied this approach to human proximal tubular cells (PTCs) in collaboration with Dr. Daniele Zink from the Institute of Bioengineering and Nanotechnology (IBN, A*STAR), and built the first high-throughput in vitro model for nephrotoxicity prediction.


Toxicity prediction based on inflammation and signaling markers: We have also developed predictive nephrotoxicity models based on the gene expression levels of two pro-inflammatory cytokines, namely interleukin (IL)-6 and -8. These models can accurately predict the toxicity of xenobiotics in both primary human PTCs and induced pluripotent stem cells (iPSC)-derived PTC-like cells.


Single-cell phenotypic profiling: We develop image and data analysis tools to extract biological information from microscopy images generated from high-throughput, cell-based screening experiments. These tools include the Protein Localization Analysis and Search Tools (PLAST) and the "cellXpress" software platform.

 
News & Announcements
  • (10 May 2017) We are recruiting a software engineer in big data analytics.
  • (15 Feb 2017) Our paper on predicting TNFa sensitivity based on early spatiotemporal-specific changes in intermediate signals has has been accepted by Scientific Reports.
  • (12 Nov 2016) Daniele and Lit-Hsin have won the Lush Prize 2016, Science Award for their work in developing animal-free Methods to predict the toxic effects of chemicals on the human kidney. This competitive, international prize aims to recognize research work that produces effective non-animal safety tests. The Straits Times published an article about our award. The award ceremony was held in London [online video].
  • (30 June 2016) Our review paper on image-based phenotypic profiling has been accepted by Cytometry A. Congratulations to Nic!
  • (28 May 2016) A postdoctoral research fellow position in computational toxicology is available in our lab. Please visit the "Careers" page.
  • (5 Feb 2016) Our work in nephrotoxicity prediction was featured in the Science section of today's Straits Times [Main article, Side article].
  • (1 Feb 2016) Lit-Hsin presented the work of the group at the APFP2016 Meeting in Bangkok, Thailand.

Our research


The research in Loo's Lab is focused on the development of in vitro and computational models for predicting the toxicity and/or targets of chemical compounds. We develop novel cellular phenotypic profiling methods and tools that can automatically classify the effects of compounds with diverse or even unknown chemical structures. These technologies allow us to build highly accurate and scalable in vitro cell-based models for predicting organ-specific toxicities. Our models can be used as high-throughput alternatives to animal testing. Our current projects include:



Toxicity prediction based on phenotypic profiles

Xenobiotics are chemicals foreign to the body, such as drugs, food ingredients, industrial chemicals, toxins, or environmental pollutants. Major organs in the body that are targeted by xenobiotics include the liver, kidneys, lungs, and heart. Animal testing is a standard approach for predicting xenobiotic toxicity in human, but suffers from the problems of long turnaround time, low throughput, and sometimes poor prediction of human toxicity. To overcome these problems, our group has developed an approach to predict xenobiotic toxicity using high-throughput imaging of cultured human cells, quantitative phenotypic profiling, and machine learning models. Instead of using standard cell death indicators as toxicity endpoints, we unbiasedly measure large numbers of quantitative phenotypic descriptors (or “features”) under in vitro conditions, and automatically search for feature combinations that can accurately predict in vivo organ-specific toxicity. Our models based on the identified phenotypic profiles can accurately classify xenobiotic compounds according to their modes of action without any prior knowledge of their chemical structures or mechanisms. Therefore, our approach is especially suitable for testing large numbers of chemicals with diverse or unknown structures or injury mechanisms.

We have applied this approach to human proximal tubular cells (PTCs) in collaboration with Dr. Daniele Zink from the Institute of Bioengineering and Nanotechnology (IBN, A*STAR), and built the first high-throughput in vitro model for nephrotoxicity prediction. Using phenotypic profiling, we identified a small set of chromatin and cytoskeletal features that can predict the human in vivo PTC toxicity of 44 reference compounds with ~82 % (primary PTCs) or 89 % (immortalized PTCs) balanced accuracies. Surprisingly, our results also revealed that a DNA damage response is commonly induced by different PTC toxicants that have diverse chemical structures and injury mechanisms. We also found that PTC-toxic compounds induced highly variable cell death responses, and therefore in vitro cell death indicators cannot be used to replace our phenotypic features. Together, our results show that human nephrotoxicity can be predicted with high efficiency and accuracy by combining cell-based and computational methods that are suitable for automation. Our nephrotoxicity models are ready for large-scale and industrial applications. Currently, we are also developing models for other major organs targeted by xenobiotics.


In the news:

  • (June 2016) A*STAR Research Highlight [Link].
  • (Feb 2016) Science section of Straits Times [Main article, and side article].
  • (Jab 2016) A*STAR's press release [Link].


Related publications:

  • (2015) High-throughput imaging-based nephrotoxicity prediction for xenobiotics with diverse chemical structures, Ran Su, et al., Archives of Toxicology, in press. [Link]



Toxicity prediction based on inflammation and signaling markers

Besides cell injury, toxic xenobiotics may also induce signaling or inflammatory responses in their targeted organs. We have also developed predictive nephrotoxicity models based on the gene expression levels of two pro-inflammatory cytokines, namely interleukin (IL)-6 and -8. Although the models only have two variables, we found that non-linear decision thresholds automatically determined using a machine learning algorithm called “random forest” gives higher prediction accuracy that linear thresholds determined either manually or using other algorithms. Our IL-6 and -8 models can accurately predict the toxicity of xenobiotics in both primary human PTCs and induced pluripotent stem cells (iPSC)-derived PTC-like cells. The results suggest that inflammation is a general response of PTCs to PTC-toxic compounds, and our computational methods are applicable to different sources of PTCs. The use of iPS cells will allow us to develop personalized or disease-specific cellular response models. We are also developing other predictive models based on the activations and subcellular localization patterns of intracellular signaling pathways.


Related publications:

  • (2015) Prediction of drug-induced nephrotoxicity and injury mechanisms with human induced pluripotent stem cell-derived cells and machine learning methods, Karthikeyan Kandasamy, et al., Scientific Reports, 5:12337. [Link]
  • (2014) Supervised prediction of drug-induced nephrotoxicity based on interleukin-6 and -8 expression levels, Ran Su, et al., BMC Bioinformatics, 15(Suppl 16):S16. [Link]



High-throughput phenotypic profiling methods and tools

Cellular phenotypes are observable characteristics of cells resulting from the interactions of internal and external chemical or biochemical factors. Imaging-based phenotypic screens under large numbers of experimental conditions can be used to study the influences of these factors to cellular phenotypes. Phenotypic profiling is a computational procedure to construct quantitative and compact representations of cellular phenotypes based on the cellular images collected in these screens. We have developed several computational methods for phenotypic profiling, which include the Drug-Profiling (“D-profiling”) algorithm, the Protein-localization Profiling (“P-profiling”) algorithm, and a random-forest-based recursive feature elimination algorithm. We have used the phenotypic profiles constructed using these methods to classify the effects of small molecules, annotate protein localization patterns, compare spatial and functional divergence of proteins, or predict toxicity effects of xenobiotic compounds.


Our group also develops the cellXpress platform, which is a software environment for performing phenotypic profiling and constructing computational models based on the generated profiles. The platform includes a user-friendly and highly-scalable software application called “cellXpress” for extracting phenotypic features from microscopy images of cells (http://www.cellXpress.org). The platform also has a web interface called “Protein Localization Analysis and Search Tools” (PLAST) for searching and comparing phenotypic profiles (http://plast.bii.a-star.edu.sg). The platform is designed to handle terabytes of image data and quantify millions of individual cells under different experimental conditions. It is used by most of our past and on-going projects.


Related publications:

  • (2014) Quantitative protein localization signatures reveal an association between spatial and functional divergences of proteins, Lit-Hsin Loo, et al., PLOS Computational Biology, 10(3):e1003504. [Link]
  • (2013) cellXpress: a fast and user-friendly software platform for profiling cellular phenotypes, Danai Laksameethanasan, et al., BMC Bioinformatics, 14(Suppl 16):S4. [Link]
  • (2007) Image-based multivariate profiling of drug responses from single cells, Lit-Hsin Loo, et al., Nature Methods, Vol. 4, 445-453. [Link]


Publications


11. (2017) Early spatiotemporal-specific changes in intermediate signals are predictive of cytotoxic sensitivity to TNFa and co-treatments. Lit-Hsin Loo, Nicola Bougen-Zhukov, Wei-Ling Cecilia Tan. Scientific Reports, 7:43541. [Link]
10. (2017) Large-scale image-based screening and profiling of cellular phenotypes. Nicola Bougen-Zhukov, Sheng Yang Loh, Hwee Kuan Lee, Lit-Hsin Loo. Cytometry Part A, 91A:115-125. [Link | PDF]
9. (2016) High-throughput imaging-based nephrotoxicity prediction for xenobiotics with diverse chemical structures. Ran Su, Sijing Xiong, Daniele Zink, and Lit-Hsin Loo. Archives of Toxicology, 90:2793-2808. [Link]
8. (2015) Prediction of drug-induced nephrotoxicity and injury mechanisms with human induced pluripotent stem cell-derived cells and machine learning methods. Karthikeyan Kandasamy, Jacqueline Chuah, Ran Su, Peng Huang, Kim Guan Eng, Sijing Xiong, Yao Li, Chun Siang Chia, Lit-Hsin Loo, and Daniele Zink. Scientific Reports, 5:12337. [Link]
7. (2014) Supervised prediction of drug-induced nephrotoxicity based on interleukin-6 and -8 expression levels. Ran Su, Yao Li, Daniele Zink, and Lit-Hsin Loo. BMC Bioinformatics, 15(Suppl 16):S16. [Link]
6. (2014) Quantitative protein localization signatures reveal an association between spatial and functional divergences of proteins. Lit-Hsin Loo, Danai Laksameethanasan, Yi-Ling Tung. PLOS Computational Biology, 10(3):e1003504. [Link]
5. (2013) cellXpress: a fast and user-friendly software platform for profiling cellular phenotypes. Danai Laksameethanasan, Rui Zhen Tan, Geraldine Wei-Ling Toh and Lit-Hsin Loo. BMC Bioinformatics, 14(Suppl 16):S4. [Link]
4. (2009) Heterogeneity in the physiological states and pharmacological responses of differentiating 3T3-L1 preadipocytes. Lit-Hsin Loo, Hai-Jui Lin, Dinesh K. Singh, Kathleen M. Lyons, Steven J. Altschuler and Lani F. Wu. Journal of Cell Biology, Vol. 187, No. 3, 375-384. [Link | JCB In-Focus article]
3. (2009) An approach for extensibly profiling the molecular states of cellular subpopulations. Lit-Hsin Loo, Hai-Jui Lin, Robert J. Steininger III, Yanqin Wang, Lani F. Wu and Steven J. Altschuler. Nature Methods, Vol. 6, 759-765. [Link]
2. (2007) Image-based multivariate profiling of drug responses from single cells. Lit-Hsin Loo, Lani F. Wu and Steven J. Altschuler. Nature Methods, Vol. 4, 445-453. [Link]
1. (2007) New criteria for selecting differentially expressed genes. Lit-Hsin Loo, Samuel Roberts, Leonid Hrebien, and Moshe Kam. IEEE Engineering in Medicine and Biology Magazine, Vol. 26, 17-26.

Current Team Members




NameTitle
LOO Lit HsinPrincipal investigator
James A. MILLERPostdoctoral fellow
Sreetama BASUPostdoctoral fellow
LEE Jia Ying JoeyResearch officer
Janice Goh Jia NiResearch officer (A*STAR NSS Scholar)


Loo Lit Hsin

LOO Lit Hsin, Ph.D.


Principal Investigator
Complex Cellular Phenotype Analysis Group
Bioinformatics Institute, A*STAR

Adjunct Assistant Professor
Department of Pharmacology
Yong Loo Lin School of Medicine
National University of Singapore
Email: lhloo at bii dot a-star dot edu dot sg

Lit-Hsin Loo is leading a small interdisciplinary team of scientists at BII developing in vitro and computational models for predicting the toxicity and/or targets of chemical compounds with diverse or unknown structures. His team has developed novel image-based phenotypic profiling methods and tools that led to the first high-throughput and predictive in vitro platform for nephrotoxicity prediction. Before joining BII, Dr. Loo was a postdoctoral fellow in the Bauer Center for Genomics Research at Harvard University (2005), and then in the Department of Pharmacology at the UTSW Medical Center, USA (2005-2010).

Awards:
  • (2016) Lush Prize 2016, Science Award
  • (2010) University of Texas Southwestern Medical Center Postdoctoral Award
  • (2009) Alfred Gilman Postdoctoral Award
  • (2005) Drexel Doctoral Award in Mathematical Sciences and Engineering



James

James A. MILLER, Ph.D.


Postdoctoral Research Fellow
Email: sur at bii dot a-star dot edu dot sg

James is an alumnus of Brasenose College, Oxford, graduating in 2009 from the Department of Chemistry. His Masters project was a theoretical study of phase spaces for confined matter inside carbon nanotubes. In 2015, he defended his doctoral thesis in computational nuclear physics at the University of Sheffield, and went on to postdoctoral work in the British National Nuclear Laboratory developing predictive algorithms for the treatment of highly active fission products. Now investigating biological as opposed to radiological toxins, James' work in the group focuses on image-based machine learning and cheminformatics. Through automated identification and classification of the microscopic changes in cells exposed to different stimuli, the intention is to develop novel methods of organ-specific toxicity prediction.


Loo Lab Alumni


Postdoctoral Fellows:
  1. Nicola Bougen (2013-17)
  2. Ran Su (2013-16) - now Associate Professor at Tianjin University, China
  3. Danai Laskameethanasan (2010-13)

Research Officers:
  1. Michelle Su Yu Fah (2014-17)
  2. Mark NEO Wei Xuan (2014-16)
  3. Cecilia TAN Wei Ling (2010-13)

Interns:
  1. Vanessa KEE Ting Zhen (Monash University, 2015)
  2. LEE Yin Yeng (Nanyang Technological University, 2014)
  3. Yi-Ling Irene TUNG (University of Michigan, 2013)
  4. TAN Wey Chyn (Multimedia University, 2011)




Careers @ the Loo Lab

We are always interested in postdoctoral-fellow candidates with backgrounds relevant to our work. Every year, we also take in one to two undergraduate interns to provide them with hand-on experiences in scientific research projects. Some of our interns also contributed to our published work. If you are interested to join our group, please send your CV to Dr. Lit-Hsin Loo (loolh at bii dot a-star dot edu dot sg). Specific openings in the lab are listed below:


  1. Software Engineer in Big Data Analytics (10 May 2017)
  2. Internship in cell biology
  3. Internship in computational analysis of microscopy images


Software Engineer in Big Data Analytics

Description:

A Software Engineer position in the area of big data analytics is available in the Complex Cellular Phenotype Analysis Group at the Bioinformatics Institute (BII), Agency for Science and Technology Research (A*STAR). The successful candidate will be part of an interdisciplinary team that develops novel computational models for predicting the toxicity of drugs, food ingredients, industrial chemicals, and environmental pollutants.

He/she will design and develop software packages and web interfaces for analyzing, modeling, and visualizing the big data generated from large-scale cellular imaging experiments. The work will involve programming, software and user interface design, algorithm implementation, software profiling and testing, and database administration. The candidate will have the opportunity to work in a highly stimulating scientific research environment, and participate the development of next-generation computational technologies for toxicity testing.

Qualifications:

  • MSc or BSc in computer science, computer engineering, statistics, or other related engineering disciplines,
  • Must be proficient in either C++, R, or PHP programming; and familiar with HTML/CSS, SQL, and the Linux operating system,
  • Ideally, candidate should have at least 2-3 years working experience in software engineering or development, but fresh graduates with relevant experience are also welcome to apply,
  • Past experience in machine learning, statistics, image processing, or bioinformatics is highly desirable, but not required, and
  • Strong problem-solving skills, and able to work independently.


Application Procedures:

Applicants should send a one-page cover letter briefly describing prior research/work experience and accomplishments, and a full resume with contact information for two referees to Dr. Loo Lit Hsin (loolh at bii dot a-star dot edu dot sg). Only shortlisted candidates will be notified.


Internship in Cell Biology

Description:

An internship position in the area of cell biology is available in the Loo lab at the Bioinformatics Institute (BII) in Singapore. The successful candidate will study the responses of lung epithelial cells to drugs and other chemical compounds. He/she will have the opportunity to work in a highly interdisciplinary and stimulating environment, and engage in innovative cell biology research. The preferred duration of internship is three months, and the starting date is flexible.

Qualifications:

Candidates must have completed university course work in the areas of biological and/or chemical sciences. Previous experience in performing cell culture work and/or molecular biology experiments is desired.

Application Procedures:

Applicants should contact Dr. Lit-Hsin Loo (loolh at bii dot a-star dot edu dot sg) for more information.


Internship in Computational Software Development

Description:

An internship position in the area of computational software development is available in the Loo lab at the Bioinformatics Institute (BII) in Singapore. The successful candidate will participate in the software development of image processing and machine learning tools. He/she will have the opportunity to work in a highly interdisciplinary and stimulating environment, and learn how computer science can help biologists to make biological discovery. The preferred duration of internship is three months, and the starting date is flexible.

Qualifications:

Candidates must have strong knowledge in C++ programming under the Linux environment, and basic knowledge of machine learning and web programming. Previous experience in bioinformatics or molecular and cell biology is desired, but not required. Candidates with interdisciplinary training in computer science and biology are especially encouraged to apply.

Application Procedures:

Applicants should contact Dr. Lit-Hsin Loo (loolh at bii dot a-star dot edu dot sg) for more information.

Contact us


The Loo Lab is located at:


Bioinformatics Institute
30 Biopolis Street, #07-01 Matrix,
Singapore 138671

Email: loolh at bii dot a-star dot edu dot sg
Tel: (65) 6478 8298
Fax: (65) 6478 9048


[Direction to BII]


About BII


The Bioinformatics Institute (BII) was set up by the Agency for Science and Technology Research (A*STAR) in July 2001; it was re-launched with a strong scientific program in the autumn months of 2007. Located in the Biopolis, BII is conceived as the computational biology research and postgraduate training institute as well as a national resource centre in bioinformatics within the Biomedical Research Council (BMRC) of A*STAR.

The BII focuses on theoretical approaches aimed at understanding biomolecular mechanisms that underlie biological phenomena, the development of computational methods to support this discovery process, and experimental verification of predicted molecular and cellular functions of genes and proteins with biochemical methods. Together with the BMRC, A*STAR research institutes and multinational R&D organizations in the Biopolis, the BII is situated in a conducive environment for exchange of scientific knowledge and friendly interaction that will prompt greater collaborations, and position the Biopolis as a notable biomedical R&D hub in Asia and in the world.



© 2011-17, Complex Cellular Phenotype Analysis Group.