Sydney Precision Data Science Centre
Sydney Data Science Insights
Edition 2, Dec 2023
|
|
|
|
Message from the Director - Professor Jean Yang
Dear Colleagues,
Semester 2 has been busy with events and new Centre activities getting underway. It was great to be able to attend conferences locally and internationally.
New activities for the Centre include a workshop co-presented by Dr Shila Ghazanfar and Dr Ellis Patrick about understanding spatial genomics. Following excellent feedback from this we hope to deliver similar workshops each year. I am happy to report that our inaugural Winter Data Analysis Challenge was a great success. This year we challenged students to develop a research question related to the UN’s Food and Agriculture Organization’s data. Prizes were awarded based on creativity of analysis and clarity of communication.
Many of our members have received fellowships, grants and awards this semester. Please check out the news and updates section for all the details. Congratulations to all students and staff who have received funding or awards! I was delighted to be awarded SUPRA Supervisor of the Year after being nominated by my students. Thank you to my wonderful team of students and SUPRA for the great evening at the awards ceremony.
I hope you enjoy catching up on Centre activities in this newsletter and I look forward to seeing you at our Centre events in 2024.
Best wishes,
Jean
| |
|
Resercher Spotlight: Dr Garth Tarr
In this newsletter we are spotlighting Dr Garth Tarr, Director of Research Division (Statistics) and one of our founding research leaders.
What got you interested and excited about data science?
I sort of fell into data science. There was no grand plan to become a data scientist when I left high school. As part of my undergraduate degree, I was required to do first year econometrics and that’s where I first felt it click – the hook for me is the buzz I get from discussing a new problem with collaborators, opening a fresh data set, and diving in to see what we can find.
What is keeping you busy at the moment?
The second half of my year gets consumed by teaching and coordinating DATA2002, the largest second year unit in the university. This is a labour of love. When I reflect on the skills my second year students walk away with compared to what I knew at the end of my second year they are so much more capable in terms of working with data sources, thinking critically about their data modelling, reproducible reporting, version control, team work and presentation skills.The feedback I’ve had from graduates suggests that it’s one of the most useful units they take in their studies. On the research side, my students and I bring together robust methods, regularisation, feature selection and stability, multi-level modelling and more recently some time series analyses to tackle problems arising mostly in the agricultural sciences. There’s no shortage of complex data sets that require careful analysis and the development of new techniques!
How long have you been working with the SPDS?
I’ve been with SPDS since the beginning and with Sydney Precision Bioinformatics before that. It’s the vibrant and inclusive research community across all levels (students and staff) that keeps me coming back for more!
Where could we find you when you’re not working?
Scooting or playing soccer with my child at one of the parks in Camperdown or travelling to visit friends and family scattered around Australia.
| |
Feature method: BIDCell
In this issue Dr Helen Fu shares an overview of BIDCell as our featured method.
The recent emergence of subcellular in situ spatial transcriptomic (SST) technology platforms has created the opportunity for researchers to quantify the spatial distribution of hundreds of genes at subcellular resolution, which allows biological insights into cell-cell interaction, tissue organisation, and tumour microenvironments. This, however, requires accurate cell segmentation to correctly assign detected gene expression to cells and uncover insights
We present BIDCell, a biologically informed deep learning-based cell segmentation framework for subcellular spatial transcriptomics data. BIDCell incorporates multiples loss functions that synergistically work to learn mappings between spatial gene expression and cell morphology, without requiring ground truth. Furthermore, BIDCell incorporates prior knowledge from single-cell sequencing databases to enable the estimation of different cell shapes and ensure the purity of gene expression within cell bodies
Alongside BIDCell, we also created CellSPA, which assesses five complementary categories of criteria with over 25 metrics for cell segmentation performance. BIDCell outperformed other state-of-the-art methods according to many CellSPA metrics across a variety of tissue types. The method produced segmentations that enhanced the purity of cell expression, providing a more accurate topographic context for analysing neighbouring cellular interactions. BIDCell is generalisable across different platforms such as Xenium, CosMx, MERSCOPE, and Stereo-seq, and is available as a Python package. BIDCell can empower researchers to enhance single-cell spatial expression analyses, enabling great potential in biological discovery.
Find out more in the paper on bioRxiv.
| |
Above image: Spatial image highlighting the cell type and neuronal regions of cells segmented using BIDCell on mouse brain Xenium data
| |
News & Updates from Semester 2
July
- The Centre kicked off semester 2 with a BBQ in Victoria Park hosted by the ECR Committee.
- Congratulations to Dr Alistair Senior on being awarded ARC Future Fellowship. Alistair will be investigating how genes and diet interact to determine lifespan over the lifetime of individuals and intergenerationally. This will involve developing and using the latest quantitative analysis tools.
- Several centre research leaders attended the Single-Cell Plus – Data Science Challenges in Single-Cell Research BIRS23 conference in Banff, Canada from 2-7 July, co-organised by Prof Jean Yang.
- The inaugural Winter Data Analysis Challenge ran from 7-23 July, offering a chance for undergraduate students to showcase skills and creativity in skills and creativity in analysing a complex and challenging dataset. Prizes were awarded for 1st, 2nd and 3rd place.
- Shila Ghazanfar and Ellis Patrick co-presented a four-hour workshop that focused on the newest techniques for understanding spatial genomics and spatial proteomics datasets.
August
- Congratulations to Dr Helen Fu, Prof Jinman Kim, and Prof Jean Yang on being awarded a USyd-UofG Ignition Grant for their proposal titled Enabling biological insights from histopathology images with artificial intelligence (AI) for precision medicine in cancer. This project is in collaboration with Dr Ke Yuan and Prof David Chang from the University of Glasgow.
September
- Congratulations to Harry Robertson, Dr Ellis Patrick and Prof Jean Yang on being selected for the Blackbird Foundry competition. Over the 8-week program the team will be honing a solution to a problem and then pitching their idea on the final competition day in December
- Congratulations to our PhD students and Postdoctoral Fellows who received awards at Sydney Bioinformatics Research Symposium! Yue Cao: Best Poster Presentation, Farhan Ameen: Best Fast Forward Presentation, Daniel Kim: Best Fast Forward Presentation, Andy Tran: Best Oral Presentation, Hani Kim: Best Oral Presentation.
October
- Dr Ellis Patrick and Dr Dario Strbenac co-organised the 2023 BioC Asia Conference in Hong Kong held from October 16-17.
November
- Congratulations to Prof Patrick Tam and A/Prof Pengyi Yang on being awarded the ARC Discovery Grant for their project: Assembling the building blocks in the blueprint of the embryonic head.
- Prof Jean Yang and Dr Ellis Patrick were both nominated by their students for SUPRA Supervisor of the year! Well done to Prof Yang for winning the award and Dr Patrick for being a finalist.
- On 7 November, we met up with researchers from the Drug Discovery Initiative for a half-day collaborative workshop to identify grand challenges in drug discovery and data science. The interactive session was a great chance for the two centres to learn more about each other's interests and strengths. We look forward to new and exciting collaborations between the two centres.
| |
Above images: Left - Dr Shila Ghazanfar presented certificates to winners of the Winter Data Analysis Challenge. Middle - brainstorming session at the collaborative workshop with Drug Discovery Initiative. Right - Prof Jean Yang celebrates with her students after being awarded SUPRA supervisor of the year.
| |
Wrap up of our seminar series
Dr Yue Cao coordinates the semiars and shares this summary of the latest series.
Our Statistical Bioinformatics Seminar Series is delivered each Monday during the semester and includes Judith and David Coffey invited speakers. The most recent series featured national and international speakers including visitors from the US and Canada. We covered a diverse range of topics including microbiome, kidney disease, spatial omics and variant annotation. Speakers span all career stages including PhD candidates, post-docs and professors. We’re pleased to note the gender balance was very close to 50:50 with five female and four male presenters.
Following the formal presentations, our team enjoys having a more casual conversation with the speakers where we can dive deeper into the topics and get to know one another. It was particularly intriguing to hear about the microbiome research from Professor Elaine Holmes who joined us one of the Judith and David Coffey speakers. Some of Elaine’s work has revealed unexpected positive effects in young mice receiving gut microbiota transplants from old mice. These effects include increased neurogenesis and prolongevity signaling. These unexpected findings provoke thoughts on the relationship between age and microbiome.
It was also interesting to hear Assistant Professor Ying Ma’s work on CARD, a spatial deconvolution method. The advantage of the work lies in that it does not require a single-cell RNA-seq reference data, unlike many of existing methods, therefore the deconvolution is not limited to the cell types in the reference data.
Looking ahead to the upcoming year, we will be continuing the seminar series starting from late February/early March. Many of our seminars are recorded and available for centre members. If you would like to view the seminar videos please email data-science.admin@sydney.edu.au
Please keep an eye on our website and stay on our mailing list to find out about upcoming events.
| |
ECR Activities
The Centre’s ECR committee led by Dr Helen Fu, Daniel Kim, Daniel Mechtersheimer, and Farhan Ameen delivered a range of activities in the second half of 2023. Activities were enabled by the centre and supported jointly by research leaders across schools. Initiatives included:
- Cross-clusters BBQ (21 July). Around 45 student participants came together to network with members of different clusters.
- ECR retreat at Mandalong (26-28 July). 12 students and ECRs attended the 3-day retreat. The participants enjoyed group activities including hiking, BBQs, and sightseeing along the Central Coast.
- Hackathon (26 Sept). A successful event with 26 attendees working on a challenge to perform bioinformatics analysis and communicate their findings, through storytelling and informative visualisations in an accessible and inclusive way. Prizes were awarded to 4 teams.
- Group dinner (9 Nov). Delicious dinner at Bornga KBBQ Restaurant to farewell A/Prof David Lin.
| |
Above images: Left - Cross cluster BBQ organised by the ECR committee. Centre - Dinner at the ECR retreat house on the Central Coast. Right - Hackathon participants deep in problem solving.
| |
|
Centre visitor A/Prof David LinGood friend and mentor of many members of SPDS, Associate Professor David Lin visited Sydney in October. David is based at Cornell and spent two weeks at SPDS. In this article, HDR Student Farhan Ameen shares his thoughts on the benefits of visitors like David.
Storytelling in science is an art, which transforms complex and abstract concepts into clear yet engaging narratives. Without a compelling story, we limit the accessibility and reach of crucial scientific concepts within the broader community. My thoughts often get lost in translation when writing, especially in our field where it’s easy to get bogged down in technical details. This challenge was strongly felt in my current project, where the “story” never felt quite right. What started off as a simple research pitch to Dave, grew into an exchange of perspectives and ideas. We began to explore and reframe the problem guided by Dave’s unique insight. I was astonished by his ability to not only grasp a completely foreign concept but to distill and weave it effortlessly into the broader scientific landscape. Viewing my work through Dave’s lens, renewed a sense of confidence in me. I started to recognise my work's potential and impact beyond just my technical niche. These experiences underscore the value of collaboration and storytelling in the scientific process. They stand as powerful tools for inspiring future scientists, fostering a culture of curiosity, and broaden the reach of science to create a more inclusive and engaging experience for everyone. Thank you for your curiosity and insight Dave!
Image: Prof Jean Yang (left) farewells A/Prof David Lin (right) with students and post-docs over dinner.
| |
|
|
|
|
|
