Tracing Early Trading in Singapore Using Geochemical Fingerprints and Artificial Intelligence on Archaeological Artefacts
PI: Ivy Yeh Hui-Yuan (SOH)
Co-PI: Wang Xianfeng (ASE); Wen Yonggang (SCSE)
Singapore is celebrating its Bicentennial as a trading hub in Southeast Asia. History of the early trading in Singapore and adjacent regions however remains fragmentary in written accounts. Archaeological artefacts discovered in the region can provide direct evidence of commodity sources, trading routes and early settlements. The current study on artefacts typically adopts a comparative method, which is limited by artefact preservations, and artefact identifications are often time consuming, qualitative and even ambiguous. Here, we propose to answer the question of how the early populations traded in Singapore and Southeast Asia by employing geochemical fingerprints to trace the geological origin of the archaeological artefacts and Artificial Intelligence (AI) technology for image analysis of material patterns. Trace metal concentrations and isotope ratios in archaeological artefacts can help trace the origin of the materials, and smelting and kiln processes. In addition, image analysis with AI technology can identity artefact patterns with dramatically high efficiency and accuracy. Both methods have not yet been employed to study archaeological artefacts in Singapore and her neighbouring areas, where hundred thousands of sherds have been excavated. By applying the two approaches, our study can potentially reveal a new chapter of the trading history of Singapore. Our proposed study is multi-discipline (e.g., archaeology, geochemistry, and computer science), and cannot fit in a typical application category of MOE AcRF Tier-1 grant. Our project will also be adventurous and innovative, and cannot fit either to the grant calls of the National Heritage Board (NHB) which typically support modern historical studies of post-19thcentury. Nevertheless, our study can provide a powerful tool to advance joint researches between archaeologists and scientists, and timely shed light on the important history of early trading and settlements in Singapore and Southeast Asia.
PI: Alper Darendeli (NBS)
Co-PI: Tay Wee Peng (EEE);Sun Aixin (SCSE)
The massive spread of fake news and misinformation is considered a major global risk with the potential to influence elections and threaten democracies. Social media is the main conduit through which false news and rumours propagate and facilitate disinformation campaigns in the political sphere. While the recent U.S. election puts spotlights on fake news’ political ramifications, little work has been done on the negative externalities of fake news on overall economy. Our study will be one of the first exploring potential implications of fake news for overall economy and capital markets. We will try to map the geographical origin of fake news and provide initial evidence on the potential use of fake news by foreign actors on countries’ economic matters.
The nature of the research question requires tools and techniques from different academic disciplines, such as computer science, business and electrical & electronic engineering. The ACE award will enable collaboration of faculty from different subject areas by facilitating the inter-disciplinary use of methodologies developed within their usual domain-constraints. ;
Artificial Intelligence for the Prediction of Alternative Splicing from Epigenomics and Transcriptomics Data in Cancer
PI: Melissa Fullwood (SBS)
Co-PI: Kwoh Chee Keong (SCSE); Francesc Xavier Roca Castella (SBS)
Collaborator: Kelin Xia (SPMS)
Many epigenetic processes, such as DNA methylation, are frequently dysregulated in cancer, which has led to the development of epigenetic drugs that are used in cancer treatment. The recent recognition that alternative splicing is largely co-transcriptional suggests that epigenetic factors may control alternative splicing, raising the possibility of using epigenetic drugs to correct alternative splicing defects in cancer, or of identifying unintended effects on splicing as a result of epigenetic drug usage in the clinic. Artificial intelligence, such as deep learning methods for making predictions from genomic data, could be used to better understand the relationships between epigenetics and splicing which will be necessary in order to develop safer and more effective epigenetic therapies for cancer.
Our hypothesis is that epigenomic events such as histone modifications, transcription factor binding, and 3D genomic organization (chromatin interactions) as well as splicing junctions will predict alternative splicing events from open chromatin data with high accuracy. We will develop deep learning models of alternative splicing events using common cancer cell lines such as K562, GM12878 and MCF-7, and test our models using RNA-Seq data from the same cells. Next, we will use integrated datasets from patient samples such as leukemia samples. Finally, we will investigate whether we can predict changes in alternative splicing from changes in open chromatin after treating cells with epigenetic inhibitors, such as methylation inhibitors decitabine and azacytidine.