[ Multi-omic Analysis ] Immune Response Against COVID-19: an -omics approach

 

여러 -omic 데이터를 통합하여 (more data, potential to more power) 우리는 complex biological big data 를 분석할 수 있다. 

( Genomics, Transcriptomics, Proteomics, Epigenomics )

* Epigenome: DNA나 히스톤단백질에 직접 결합하여 유전자의 발현을 직접 조절하는 화학물질과 단백질의 집합체

 

 

 Abstract

To understand the human responses to the virus (COVID-19), especially in the Immune response. 따라서, 해당 페이퍼의 목표는 COVID-19로인한 면역 반응에서 큰 역할을 하는 immune genes를 발견하는 것이다.

 

사용한 Multiomic Datasets:

• Microarray (E-MTAB-8871)
• RNA-seq (E-MTAB-9221)
• RNA-seq (GSE152418)
• Chip-seq (GSE108881)

 

 Introduction 

면역반응은 [1] innate immune response [2] adaptive immune system로 나뉠 수 있다. 반응은 T cell 과 B cell 에 의해 일어나고 T Cell에는 CD4과 CD8 T cell (NK) 이 있다. CD4 T cell (helper T cell)은 cytokines을 분비해서, 면역 반응을 불러온다. 하지만 해당 과정에서 cytokine이 대량 분비시 문제가 될 수 있다. 

 

# Used -Omics Data 

Microarray	# Whole Blood Cells: Erythrocytes, Leukocytes, and platelets  - Transcriptomics profiles of blood sampled via NanoString Human Immunology V2 Panel  - Negative and positive controls in probe sets(for normalization)  - Time series samples
RNA-seq (PBMC samples)	# Peripheral Blood Mononuclear Cells (PBMCs): Lymphocytes  - Sequencing w/ a single Illumina HiSeq2000 flow cell  - Covid patients in four stages (convalescent, moderate, severe, and intensive care unit)
RNA-seq (organoid)	# Organoids (Lung Epithelial cells) infected with COVID  - Sequencing w/ Illumina NovaSeq 6000
ChiP-seq	# Calu3 (human lung cancer cell) infected w/ MERS-Cov (middle east respiratory syndrome coronavirus)  - PMS (peripheral blood smear) sonicated and immunoprecipitated w/ anti-H3K4me,  anti-H3K27me3    * H3kme3: epigenomic modification on Histon H3 for gene expression regulation (발현 촉진)    * H3K27me3: associated with downregulation (발현 억제)

* No nuclei: Erythrocytes, platelets 

* Mononuclei: Lymphocytes (T cells, B cells, NK cells)

* Multi-lobed nuclei: Granulocyte ( Neutrophils,basophils, eosinophils)

* White blood cell: Blood cells (except Erythrocytes or platelets)

 

 

 Analysis Methods  

# Used R-packages and Analysis Process 

	Preprocessing & Expression Analysis	Gene Set analysis
Microarray	NanoStringR(nanoR)  [0] rcc files  [1] QC, background correction, normalization       * background correction by SD       * geometric mean for positive control normalization       * housekeeping normalization  [2] time-series categorization   limma  [3] Differential Expression Analysis   [4] FDR Correction (Benjamini-Hochberg)	goana (limma)  - to determine Gene Ontology term  kegga (limma)  - to find over-represented pathways in DEG
RNA-seq (PBMC samples)	STAR  [0] Reads, mapped to human genome (GRCH38)         - counted w/ STAR using htseq-count   EdgeR  [1] DGE       - Normalization       - DE through a quasi-likelihood F-test  [2] FDR Correction (Benjamini-Hochberg)	"
RNA-seq (organoid)	SRA toolkit   [1] SRA (Sequence Read Archive) --> Fastq  [2] QC, trimming (Trimmomatic)  [3] Aligned onto GRCH38 (Kallisto)   EdgeR  [4] Count Normalization  [5] Statistical analysis	"
ChiP-seq	SRA toolkit   [1] SRA (Sequence Read Archive) --> Fastq  [2] QC, trimming (Trimmomatic)  [3] Aligned onto GRCH38 (STAR)  [4] Peak Callings (MACS2)  [5] MACS2 broad peak file --> GRanges  [6] Identify the overlap peaks w/ H3K27me3/H3K4me3	"

* Type 1 error (FDR: False Discovery rate) Correction    #  틀렸는데 맞다고 판단 

[1] Benjamini-Hochberg (BH)  https://www.youtube.com/watch?v=K8LQSvtjcEo&t=633s 

[2] Bonferroni  ( original p-vale/ # of test performed)

 

* GRCH38: Genome Reference Consortium Human Build 38

* htseq-count: counts for each gene how many aligned reads overlap its exons

* Ensembl ID 또는 Entrez ID를 알고 있을 때, 그와 관련된 정보들을 org.Hs.eg.db를 통해 annotation 할 수 있다.

 

# Used Visualization

plotPCA/ plotMDS	Shows correlation or clustering by dimensionality-reduction
volcanoPlot	scatter plot that shows  statistical significance (P-value) vs FC
plotBCV (edgeR)	Shows estimate Tagwise, Common, Trend dispersions    * Tagwise: allow for a different value for the dispersion to be used for each gene
MA plot	log FoldChange vs Average Expression Level
pheatmpap	Shows the magnitude of a phenomenon as color

 

 

 Conclusion 

각각의 데이터에서는 HLA, Immune-related genes (immunoglobulin fragments (IgG receptor, IgA, IgM), B cell receptor, interleukin gene) 같은 protein-coding genes 들이 expressed more. 전체적인 결과는 다음과 같다.

[1] 모든 데이터셋에서 공통적으로 의미 있는 gene = Cytokine (HLA-DPA1,  PTGER4, NFIL3)

[2] gene set analysis 를 통해 감염 환자들에게서 mitochondria 와 oxidative phosphorylation 가 바뀐 점이 확인되었다. 

네개의 데이터가 not that compatible 이여서 생각만큼 powerful 한 결과를 얻지 못했지만 좀더 연관성있는 데이터들을 이용한다면 보다 크게 more data, more powerful result 를 얻을 수 있을 것이다. 

* 아쉬운점: 해당 프로젝트는 COVID-19 가 발생한지 (2019.12) 일년도 채 되지 않았을 때 (2020.10) 진행하였기에 충분한 데이터가 부족하였다.