Soroor Hediyeh-zadeh
I am a PhD student in Machine Learning and Computational Biology at Helmholtz Munich and the Technical University of Munich advised by Fabian Theis. My research interests are modeling the mechanisms of cell state transitions and cell state inference in different biological data modalities using Generative, Causal Representation Learning and Continual Learning models.
Before coming to Munich, I was a Graduate Student Researcher and Research Assistant in Bioinformatics at WEHI. I received a MPhil in Computational Biology from the University of Melbourne advised by Melissa Davis and Andrew Webb, where I developed statistical machine learning and deep learning models for Mass Spectrometry data. I was lucky to have the opportunity of working with Gordon Smyth during my Master's degree on incorporating transcript-level differential expression in edgeR.
My MPhil research was supported by a Australian Postgraduate Award. During PhD, I received a German Academic Exchange (DAAD) scholarship to conduct research abroad.
In addition to research, I am passionate about education and leardership. I was a co-founder of R-Ladies Melbourne, where I co-organised and delivered multiple R programming workshops. I am a co-founder of STEM-Tisch at Helmholtz Munich, where we run Inspiring Women Series talks, panel discussions and training workshops to promote diversity, specifically women's presence, in scientific positions. Outside of research, you'll find me playing or composing with my D Kurd Handpan, probably in nature.
News
- 08/2024: Check out new preprint Multimodal weakly supervised learning to identify disease-specific changes in single-cell atlases for using Multiple Instance Learning and Attention mechanism for detection of disease cell states.
- 06/2024: msImpute was highlighted in Molecular and Cellular Proteomics's Special Issue as an outstanding article!
- 03/2024: First PhD work "Disentanglement via mechanism sparsity by replaying realizations of the past" was accepted at MLGenX @ ICLR2024 [link].
- 02/2024: Check out GraphCompass: we combine geometric data analysis with statistical tests for differential expression analysis in spatial transcriptomic data!
- 11/2023: Our work on modeling quantification uncertainty in differential transcript expression was accepted in NAR!
- 10/2023: Our work scPoli is out in Nature Methods. scPoli also recieved the Helmholtz Munich Best Paper Award!
- 10/2023: New Preprint Alert! Check out scDECAF, learning vector representation of gene sets for fast query of gene programs in single cell transcriptomic data!
- 04/2023: msImpute is out in Molecular and Cellular Proteomics. Check out the latest tutorials. Find the software on Bioconductor.
- 03/2023: Watch the recording of my talk in ContinualAI seminar series here.
- 11/2022: Officially graduated from The University of Melbourne with a MPhil degree in Computational Biology. In my thesis, I developed machine- and deep learning algorithms for (label-free) mass spectrometry proteomics.
- 05/2022: Our workshop paper on Continual Learning for single cell biology accepted at 2022 ICML Workshop in Computational Biology!
- 9/2021: I moved to Helmholtz Munich to do a PhD!
- 08/2021: I presented msImpute.v2 at BioC 2021. Find the poster here. Update - msImpute.v2 got an Invited Talk in BioC Asia 2021. Stay tuned for the details!
- 06/2021: Our works on learning geneset representations in single cell RNAseq and deep representation learning of TIMS-MS1 features in mass spec are accepted at ICML 2021 Computational Biology Workshop. Come and see us at poster session!