Someone confronted on the spot, and they said “ Maybe there is one, maybe they are common, who knows what. I hope it was an outlier.” Even this explanation is full of implicit racial bias. See the full conv:

Remember this habitual offender of racial discrimination. Shame of @MIT 🙃 Hallucination is not a problem of AI, it’s a problem of this group of humans, no? Once we eliminate their corpus from the training set we might solve the hallucination problem:)

I'm shocked to see racism happening in academia again, at the best AI conference @NeurIPSConf. Targeting specific ethnic groups to describe misconduct is inappropriate and unacceptable. @NeurIPSConf must take a stand. We call on Rosalind Picard @MIT @medialab to retract and

Excited to share our latest publication in single-cell clustering! Check out our paper on "Ensemble deep learning of embeddings for clustering multimodal single-cell omics data" (Now Published!) at

Super congrats to Wendy 🎉

K-Means Clustering https://t.co/dgV9ulStUb

Sorry, I'm late. A new paper of clustering multimodal single cell data is posted at bioRxiv. A huge thanks to @ChunleiLiu0 and my amazing supervisor @PengyiYang82 and @jeanyang21 for the their support and guidance over the course of this project.

My first preprint is out on bioRxiv!✨ The identification of genes that vary across spatial domains in tissues/cells is an essential step in spatial transcriptomics data analysis. Given this, we evaluate the performance of various proposed SVG methods.

How well do “mini-organs” in a dish recapitulate the human tissue? We set out to answer this question for the human eye, a super fascinating organ! I'm excited to share our preprint on @biorxivpreprint. Check out the full story through the link! 🧵 https://t.co/3501IKJdJB

Excited to share our work @jeanyang21 @PengyiYang82 SimBench @NatureComms, that evaluates #SingleCell simulation methods.

Very excited that our latest work scJoint integrates atlas-scale single-cell RNA-seq and ATAC-seq data with transfer learning is finally out @NatureBiotech! https://t.co/zFK1pkjDHx It is a joint work and a great collaboration between Wong Lab @Stanford and @sydneybioinfo

My final project of my PhD is out in @NatureComputSci!! https://t.co/VDlMj2svdr #singlecell #cellularidentity #markergenes #differentialstability #Cepo see thread 🧵

Very excited to share my first first-author paper in @OUPBioinfo !! https://t.co/kv4hPmgkOF Please see thread. #SPS #phosphoproteome #splicesome #cancermutation

8/8 We thank our amazing reviewers and editors who helped substantially improve the quality of this work! We will continue benchmarking new methods and new updates on old methods.

7/8 This work would not be possible without my wonderful colleagues at @sydneybioinfo, @SydneyMaths, @CMRI_AUS, @cpc_usyd, @Sydney_Uni. In particular, my partner in SydneyMaths @YueCao16 and our amazing supervisors @PengyiYang82 and @jeanyang21.

6/8 Source code of scCCESS, our proposed stability-based approach for estimating the number of cell types from scRNA-seq data, can be accessed from

5/8 We summarised these results into a multi-aspect recommendation to the users and expect this work will foster future development of scRNA-seq data clustering methods.

4/8 Notably, most clustering algorithms tend to under-estimate the number of cell types when the true number of cell types increases in the datasets, whereas interestingly SC3 and Seurat to a lesser degree over-estimate when the true number of cell types increases.

3/8 We found many clustering methods have high instability in estimating the number of cell types. Besides, a higher cell clustering concordance does not necessarily mean a more accurate number of cell type estimation.

2/8 In this study, we set up a systematic benchmark framework for evaluating clustering algorithms on estimating the number of cell types from scRNA-seq data. We design 4 settings by two key aspects: the number of cell types and the cell numbers per cell type.