Project Experience

Ph.D thesis projects

Implications of potential T-cell epitopes in human respiratory virus major surface proteins for vaccine design

RSV T-cell epitope prediction for vaccine design: We utilized immunoinformatic toolkits to predict T-cell epitopes in RSV major surface proteins across different strains.

T-cell epitope diversity visulization and comparasion: I develop a approach to get T-cell epitope profile clustering based on MDS and wrap it into a toolbox via snakemake. ([GitHub]

T_epi_mds

Modeling the interaction of RSV and influenza cocirculation

](https://github.com/JianiC/RSV_flu)

This is a infectious disease modeling study which incoperate Markov process and optimization algorithms. I aim to test whether the inhibition of co-infection happens during the infectious process ( inhibit the pathogen to enter to the cell) or caused by cross-protective immunity.

pomp_model_simulate pomp_modelfit

A novel nomenclature system for RSV genotyping

First I build a novel nomenclature for RSV using phylogenetic approach. Then the sequences with assiged genotype were used a a trainging data to develop a software to perform automatic genotype classification. Support Vector Machine (SVM) model and machine learning algorithms were used.

genotype

Other projects I have contributed to

Origin of SARS-CoV19: phylogentic and contact network analysis

The goal of this project to tract the source of SARS-CoV2, phylogenetic and network approach are used.

origin of SARS

Avian influnza in Atlanta flyway: Bayesian phylogenography analysis

Discrete trait analysis and Markov jumps and rewards analyis in BEAST.

Screen Shot 2021-07-23 at 11 48 28 PM

APP: Perdict the geo-location information of SARS-CoV2 using genetic information : NLP and Naive Bayes Classifier

In order to the isolate location of the sequences with missing information, first I vectorized RNA sequences into a format that machine learning algorithms can understand using sklearn.feature_extraction.text. Then I trained a machine learning model to learn to isolated country between RNA sequences. Finally, with the trained model, the sequences with missing geo information can be perdicted with an accuracy of 93%.