Date: Friday, 25 March 2022

Time: 5:00 p.m. - 6:00 p.m.

Via Zoom: https://hku.zoom.us/j/98033706013?pwd=eTVydDJvM0thT0gyRzg2V3pGcjJQdz09

Meeting ID: 980 3370 6013

Password: 465141

5:00 p.m.

Presenter: Miss Zhuoxuan LI , PhD candidate
Primary Supervisor: Prof. Pengtao LIU
Presentation Title: Cellular interaction Detection using Moran’s I in SPATIAL transcriptomics (DM-SPATIAL)[YH1]
Abstract: Intercellular communication is crucial to multiple biomedical processes, and the advancement of single-cell transcriptomics makes high-throughput intercellular communication analysis possible. Recently, high-quality curation of ligand-receptor pairs further allows deciphering the detailed interaction between cell types from scRNA-seq data, including CellChat and CellPhoneDB. However, all current methods for ligand-receptor interaction are designed for scRNA-seq data, but cannot utilize the emerging spatial transcriptomics data, even though spatially-colocalized cells are more likely to communicate. On the other hand, a few attempts on spatial interaction analysis do not support the ligand-receptor information. Here, we are developing a statistical method to dissect cell communication via detailed ligand-receptor pairs in a spatial context. Uniquely, this method can effectively identify the ligand-receptor pairs that have positive spatial correlation and indicate the regional cell-cell interactions. We validated the model in simulated data and an intestine dataset, obtaining numerous biological insights.

5:30 p.m.

Presenter: Mr. Sheng XU, PhD candidate
Primary Supervisor: Dr. Joshua Wing Kei HO
Presentation Title: Robust microbial mutation discovery in shotgun metagenomic data using MetaQuad
Abstract: An emerging area of research in gut microbiome biology is the discovery and analysis of bacterial strains using shotgun whole genome metagenomic data via analysis of bacterial single nucleotide polymorphisms (SNPs). However, SNP calling in metagenomics data is challenging due to the low and variable coverage of sequencing data in most gut microbiome studies. Existing metagenomic variant calling tools potentially identify a large amount of false positives. We reason that co-calling of mutations across many samples can distinguish true positive from false positive SNPs. Here, we present a computational method, called MetaQuad, that uses a mixture modeling approach to identify informative mutations in a collection of metagenomic samples. Evaluation using simulated and real metagenomic data suggest that MetaQuad can reduce false positive SNPs without greatly affecting true positive rate.

ALL ARE WELCOME

Should you have any enquiries, please feel free to contact Miss Cynthia Cheung at 3917 9748.