Bokai Zhu
@ZhuBokai
Followers
70
Following
14
Media
9
Statuses
17
Joined December 2021
Check out our recent tool using deep learning to better discover immune cells in tissues! Always a pleasure to work with this great team: @ZongmingMa, @SizunJ, @GarryPNolan, @ironmathmo (Sheng Gao), @ShuxiaoC, @shaleklab and the rest who may or may not be on this platform!.
2
6
15
Always a pleasure working with such a talented team—thrilled to see our work published in Nature Genetics today! For people interested in understanding their tissue's state deeper, try our new tool MESA!.
Excited to share our new study out in Nature Genetics @NatureGenet! We introduce MESA, a spatial omics framework that draws inspiration from ecology and integrates multiomics data to study the spatial architecture of tissues in health and disease. Building on the analogy between
0
1
10
RT @DaisyYDing: Excited to share our new study out in Nature Genetics @NatureGenet! We introduce MESA, a spatial omics framework that draws….
0
6
0
Finally online in Nature Communications @NatureComms today: Huge thanks to everyone involved in the study! @SizunJ @GarryPNolan @xavi_r_c @MikeAngeloLab @shaleklab @yeoyaoyu and others that are not on this platform lol.
Wanna dissect host-microbiome interactions in situ? Incredibly excited to share work from @BokaiZhu with @GarryPNolan, and @shaleklab @MikeAngeloLab @yeoyaoyu @xavi_r_c @hanimal725 and others
1
8
26
9🔬/ The tool is available on GitHub (still under testing) and I hope the field can find it helpful! (me when debugging:)
1
1
3
8🔬/ Meanwhile in an HCC tissue, we identified Macrophages that can be further divided into more subgroups, with an interesting subpopulation that has close interaction with the tumor communities.
1
0
3
7🔬/ The method also can help pin-point important keystone populations directly related to disease states! For example, in a cHL tissue, we identified CD4T cell subpopulations that posses distinct tumor-infiltration abilities:
1
0
2
5🔬/ And yes, by fusing these cross-modality information we refine the cell populations better in spatial-comics! For example, by using CellSNAP we were able to refine distinct germinal center B cell populations in mouse spleen:
1
0
3
4🔬/ CellSNAP fuses single-cell molecular expression profiles with information like neighborhood and tissue morphology from images! To fuse these very very different types of information, we utilized GNN and novel model architectures!
1
1
5
3🔬/ These methods work great, but when adapted to spatial data, they ignore vital image-level information (and we don’t like that!). That is where our new method CellSNAP comes in, which it uses all information from spatial data!
1
0
3
2🔬/ De-novo cell population identification is a common and vital step in almost all spatial-omic studies. However, most methods utilized in this field are directly adapted from earlier modalities (eg. cytometry or single-cell sequencing). Me when annotating cells:
1
0
4
1🔬/ Excited to share our latest preprint led by @ShuxiaoC , @Shenggao10 , @GarryPNolan , @SizunJ , @ZongmingMa , and with help from @shaleklab, Scott and more!
1
25
66
RT @SizunJ: Wanna dissect host-microbiome interactions in situ? Incredibly excited to share work from @BokaiZhu with @GarryPNolan, and @sha….
0
32
0