在前面的學徒實習生數(shù)據(jù)挖掘任務列表：純R代碼實現(xiàn)ssGSEA算法評估腫瘤免疫浸潤程度 信息描述了如何下載基因集，然后使用GSVA包進行ssGSEA分析后可視化，為了考驗大家學習效果，我們布置一個新的圖表復現(xiàn)：

來自于文章：Multi-omics profiling reveals distinct microenvironment characterization and suggests immune escape mechanisms of triple-negative breast cancer 里面提到了數(shù)據(jù)：

The sequencing data is also available in GSE118527 (OncoScan), GSE76250 (HTA 2.0) and SRP157974  (WES and RNAseq)

主要是使用RNA-seq和HTA2.0芯片的表達數(shù)據(jù)，根據(jù)ssGSEA的結果對樣本進行分組，然后講故事這樣的分組的意義。

TNBC分型歷史

看到很多媒體宣傳最難治的乳腺癌有望獲得分類治療，但是其實TNBC分子分型的研究不少了，復旦大學邵志敏團隊2019發(fā)表的這個中國人TNBC隊列既不是第一個，也不會是最后一個。

• 首先是2011的meta分析，把TNBC分成6類：Basal-like 1 (BL1), basal-like 2 (BL2), immunomodulatory (IM), mesenchymal (M), mesenchymal stem-like (MSL) and luminal androgen receptor (LAR)

• 然后同樣的作者2016年在plos one 發(fā)文重新修訂了 之前的分類，變成4類：(TNBCtype-4) tumor-specific subtypes (BL1, BL2, M and LAR)

• 發(fā)表在Clin Cancer Res 2015 ，貝勒醫(yī)學院研究小組的 Burstein  等人對自己的數(shù)據(jù)，198個TNBC病人芯片表達矩陣，使用80個核心基因進行分組，得到4個TNBC的亞型。

• 發(fā)表在 Breast Cancer Research (2015) ：Gene-expression molecular subtyping of triple-negative breast cancer tumours: importance of immune response，數(shù)據(jù)在 GSE58812,  法國研究團隊的等人使用 適應性的Fuzzy-clustering 把107個TNBC 患者分成3類。

使用ssGSEA算法對CIBERSORT的免疫基因集進行分析

本文使用的數(shù)據(jù)在 GSE76250 可以下載，分析流程如下：

發(fā)表在：Nat Commun. 2019 Apr，是中國肺癌研究領域比較出名的吳一龍課題組文章圖表：

使用ssGSEA算法計算26 immune cell types比例

這26個基因集來源于文章 Immunity. 2013 Oct ， 分類如下；

• 11個是adaptive immunity

• 12個是 for innate immunity

• 3個是 for MDSC，angiogenesis, and antigen presentation machinery

使用GSVA包的ssGSEA算法，對z-score后的RNA-seq表達矩陣進行分析。有趣的是作者提供了RPKM矩陣哦，The RNA-seq FPKM data have been deposited at figshare (https:///10.6084/m9.figshare.7306364.v1).  所以理論上可以重現(xiàn)作者的分析。

可以把病人分成3組不同的免疫狀態(tài)，主要是看 IFNG, PD-L1, PD-1, and CD8 基因的表達

分型具有生存效果

RNA-seq和HTA2.0芯片的表達數(shù)據(jù)的比較

這里使用ComBat算法抹去兩個平臺的差異

在TNBC隊列驗證

同樣也是分成3類：

在METABRIC隊列驗證

也可以區(qū)分成為3類，圖片在文章里面的附件！

附件圖片

• Supplementary Figure 1. Workflow of our research.

• Supplementary Figure 2. Estimation of the optimal clustering numbers of triple-negative breast cancer microenvironment phenotypes.

• Supplementary Figure 3. Validation of microenvironment phenotypes clustering in METABRIC cohort.

• Supplementary Figure 4. Validation of microenvironment phenotypes clustering in TCGA cohort.

• Supplementary Figure 5. Comparison of potential molecules involved in the initiation of innate immunity among microenvironment clusters in FUSCCTNBC cohort.

• Supplementary Figure 6. SNV and indel neoantigen load of the three microenvironment clusters in triple-negative breast cancer.

• Supplementary Figure 7. Chromosome instability of the three microenvironment clusters in triple-negative breast cancer.

• Supplementary Figure 8. Cancer testis antigen landscape of triple-negative breast cancer.

• Supplementary Figure 9. Gene set enrichment analysis of enriched pathways in each cluster.

• Supplementary Figure 10. Batch effect evaluation after 'Combat' of RNA-seq and HTA microarray datasets.

• Supplementary Figure 11. Process and validation of mRNA clustering.

附件表格

• Supplementary Table 1. The compendium of microenvironment cell subtypes in triple-negative breast cancer.

• Supplementary Table 2. Correlation of estimated microenvironment cell numbers between our compendium and CIBERSORT or MCP-counter.

• Supplementary Table 3. Clinicopathological characteristics of three microenvironment phenotypes in FUSCC, METABRIC and TCGA cohort.

• Supplementary Table 4. Prognostic value of each cell subset by univariate Cox proportional hazards model for relapse free survival.

• Supplementary Table 5. The signatures of ten oncogenic pathways.

• Supplementary Table 6. Comparison of gene mutation frequency among clusters.

• Supplementary Table 7. Comparison of somatic copy number alterations among clusters.

• Supplementary Table 8. GO and KEGG annotation of genes in cluster-specific copy number variation peaks.