Dr Jing Li's Lab

·CURRENT STATUS:  paper finished, ready to submit

·DESCRIPTION:  Plenty of cells in proteomics spectral count expression matrices contain zero counts, and existing methods in differential expression analysis are not capable of handling the excess number of zeros. We proposed a biological network module-based zero-augmented generalized linear model (ZAGLM) for differential expression analysis of spectral count data. This ZAGLM utilizes two kinds of distribution model, one is called zero-inflated (ZI) distribution while the other are usually called hurdle distribution.

·PARTICIPATION:  lead the whole process

Fig. simulation pipeline for model assessment

Fig. results in-depth analysis

·SUMMARY:  During the whole participation, I've learned a lot skills:

1. regression
2. methodological comparison principle
3. parallel computation
4. mixture model
5. zero-inflated model
6. statistical test
7. Gene Set Enrichment Analysis (GSEA)

Besides from the skill, I also learned important things about scientific research under the guidance of Prof. Li. It's astounding how easily a researcher can deviate from the primary pivot during a study, and a good advisor can correct you timely when you get lost in nowhere. And all matters, whether important or trivial!

·REFERENCE:  

1. Biological network module-based model for the analysis of differential expression in shotgun proteomics. Xu J, Wang L, Li J. J Proteome Res. 2014 Dec
2. Network module-based model in the differential expression analysis for RNA-seq. Lei M, Xu J, Huang LC. Bioinformatics. 2017 Sep
3. Regression Models for Count Data in R. Zeileis A, Kleiber C, Jackman S. J Stat. Soft. 2008