RNA 23. Risk factor association diagram of Cox model of expressed genes in SCI articles (ggrisk)

Click to follow , Huanfeng gene

Huanfeng gene

Bioinformatics analysis ,SCI Writing articles and learning basic knowledge of biological information ：R Language learning ,perl Basic programming ,linux System commands ,Python Meet a better you

130 Original content

official account

The tutorial of Huanfeng gene not only teaches you how to use , We will also analyze some related articles regularly , Learning tutorials is just the foundation , But if the analysis results are integrated into the article, it is the purpose , I think these tutorials are not bad , And you analyzed the good results according to our tutorial and sent an article. Remember to inform us , And thank us in the article ！

 The English name of the company ：Kyoho Gene Technology (Beijing) Co.,Ltd.

Huanfeng gene official account launched a course of transcriptome analysis and clinical prediction model , Teachers who need students' letters can contact us ！

First, take a look at the transcription analysis tutorial, which is summarized as follows ：

RNA 1. Gene expression things – be based on GEO
RNA 2. SCI The article is based on GEO Of differentially expressed genes limma
RNA 3. SCI The article is based on T CGA Differentially expressed genes DESeq2
RNA 4. SCI The article is based on TCGA Differential expression edgeR
RNA 5. SCI Differential gene expression in the article MA chart
RNA 6. Differential gene expression -- Volcanic map (volcano)
RNA 7. SCI Gene expression in the article —— Principal component analysis (PCA)
RNA 8. SCI Differential gene expression in the article – Thermogram (heatmap)
RNA 9. SCI Gene expression in the article GO notes
RNA 10. SCI Gene expression is enriched in the article –KEGG
RNA 11. SCI Gene expression is enriched in the article GSEA
RNA 12. SCI The calculation method of tumor immune infiltration in this article CIBERSORT
RNA 13. SCI Of differentially expressed genes in the article WGCNA
RNA 14. SCI Of differentially expressed genes in the article Protein protein interaction network (PPI)
RNA 15. SCI One of the fusion genes in the article FusionGDB2
RNA 16. SCI The visualization of fusion genes in this article
RNA 17. SCI Screening in the article Hub gene (Hub genes)
RNA 18. SCI Analysis of gene set variation in the article GSVA
RNA 19. SCI Unsupervised clustering method in the article （ConsensusClusterPlus）
RNA 20. SCI Single sample immune infiltration analysis in the article （ssGSEA）
RNA 21. SCI Single gene enrichment analysis in the article
RNA 22. SCI In this paper, the stromal cells and immune cells of malignant tumor tissues are estimated based on expression （ESTIMATE） original The modified
RNA 23. SCI The risk factor association diagram of the expressed gene model in the article （ggrisk）

The clinical prediction model is summarized as follows ：

Topic 1. _ clinical _ Conventional thinking of biomarker Shengxin analysis
Topic 2. Survival analysis Kaplan-Meier
Topic 3. SCI The first table of the article – Baseline table
Topic 4. _ clinical _ Prediction model construction Logistic Return to
Topic 5. Sample size determination and segmentation
Topic 6 Counting variable Poisson regression
Topic 7. _ clinical _ prediction model –Cox Return to
Topic 8. _ clinical _ prediction model -Lasso Return to
Topic 9. SCI The second table of the article — Single factor regression analysis table
Topic 10. Single factor Logistic regression analysis — Univariate analysis table
Topic 11. SCI Multivariate screening — single / Multifactor table
Topic 12 _ clinical _ prediction model — Nomograph (Nomogram)
Topic 13. _ clinical _ prediction model — Consistency index (C-index)
Topic 14. _ clinical _ Calibration curve of prediction model (Calibration curve)
Topic 15. _ clinical _ Decision curve of prediction model (DCA)
Topic 16. _ clinical _ Receiver operation characteristic curve of prediction model (ROC)
Topic 17. Missing value recognition and visualization of clinical prediction models
Topic 18. Missing value interpolation method of clinical prediction model

Yesterday I shared an article IF：4+ Iron metabolism and immune related gene markers predict clinical outcomes and molecular characteristics of triple negative breast cancer , A teacher asked questions , Say... In the article Fig.3 c,f,i How to draw the correlation diagram of risk factors ？ In fact, there are ready-made software packages , Super easy , Let's see ！！！

Preface

Risk map is probably one of the most commonly used graphs for tumor gene data analysis . We can draw the following two conclusions : Compare the prediction results of the model with the actual survival situation to observe whether the survival rate of the high-risk group is lower than that of the low-risk group , Whether the survival time is shorter than that of the low-risk group . The other is to compare the heat map and the scatter diagram , Look at the correlation between predictors and results . Risk factor correlation diagrams are common in cox Risk proportion model , It is commonly seen in the following figure, which shows the combination of three figures , It is shown by cox The proportion of high-risk and low-risk groups distinguished by the risk model 、 Time to live （ raw / die ） And gene expression of concern （ It is usually the genes involved in model construction ） Differences in distribution . Of course, when the model does not involve gene expression , Often, the risk factor correlation diagram only shows the upper two parts .

Software installation

In the absence of this ggrisk Before the bag , Drawing still requires that each diagram be drawn separately in the group diagram , There is now a ggrisk The software package is very convenient , Read in the data and you can get the result , and ABC The picture numbers are all there , Open not happy ？

if (!require(ggrisk)) install.packages("ggrisk")

library(ggrisk)
library(survival)
library(survminer)

data fetch

We use ggrisk Package built-in data set LIRI Liver cancer from Japan ICGC database , Including time 、 Events and four genes were tested .

ICGC Liver Data from Japan Description This data is a liver cancer data from Japan Data released in ICGC database (Link). It cantains time, event and four genes. An object of class data.frame with 232 rows and 6 columns.

data(LIRI)
head(LIRI)
##        time status      ANLN     CENPA     GPR182      BCO2
## 1 3.0410959      1  6.821354 3.0366550 0.00000000 0.2248344
## 2 2.5479452      0  1.073527 0.4654169 0.17895040 5.8924860
## 3 4.0273973      0  2.579530 0.7732644 0.06809686 3.5994330
## 4 0.1643836      1 14.183630 7.7239000 0.03749626 1.1194870
## 5 0.8219178      0  3.588320 2.3237710 0.16762610 2.6660850
## 6 2.8767123      0  6.079665 3.6674980 0.21788230 0.7691067
#  Construct multifactor using four genes cox The regression model ：

Example operation

1. structure COX Return to

We use four genes to construct multifactor cox The regression model ：

library(rms)
fit <- cph(Surv(time, status) ~ ANLN + CENPA + GPR182 + BCO2, LIRI)

2. Risk factor linkage map drawing

ggrisk(fit, cutoff.value = "median", cutoff.x = 145, cutoff.y = -0.8)

3. Adjust the risk score cutoff And position

ggrisk(fit,
       cutoff.value='cutoff', # Optional ‘median’, ’roc’ or ’cutoff’
       cutoff.x = 150,  #“cutoff” The horizontal position of the text 
       cutoff.y = -1#“cutoff” The vertical position of the text 
)
##    -- 
##    ==
##    Combination:  185 
##    Combination:  43

4. Other parameter settings

Other parameter settings include color , label , Modification of drawing elements such as legend , Specific instructions can be used directly ?ggrisk Check usage .

ggrisk(fit,
       code.highrisk = 'High Risk',# High risk label , The default is  ’High’
       code.lowrisk = 'Low Risk', # Low risk label , The default is  ’Low’
       title.A.ylab='Risk Score', #A chart  y Axis name 
       title.B.ylab='Survival Time(year)', #B chart  y Axis name , Pay attention to distinguish between year month day
       title.A.legend='Risk Group', #A Figure legend name 
       title.B.legend='Status',     #B Figure legend name 
       title.C.legend='Expression', #C Figure legend name 
       relative_heights=c(0.1,0.1,0.01,0.15), #A、B、 Heat map annotation and heat map C The relative height of     
       color.A=c(low='green',high='red'),#A The color of the midpoint in the figure 
       color.B=c(code.0='green',code.1='red'), #B The color of the midpoint in the figure 
       color.C=c(low='green',median='white',high='red'), #C The color of the heat map 
       vjust.A.ylab=1, #A In the figure y Axis label to y Distance of coordinate axis , The default is 1
       vjust.B.ylab=2#B In the figure y Axis label to y Distance of coordinate axis , The default is 2
)

5. Assign heat maps to show genes

We can also specify which genes to display the heat map , as follows ：

ggrisk(fit, heatmap.genes = c("GPR182", "CENPA", "BCO2"))

6. Don't show the heat map

Show only the scatter chart , Don't show the heat map , Use it directly two_scatter() function , as follows ：

two_scatter(fit, cutoff.value = "median", cutoff.x = 142, cutoff.y = -0.5)

Is this very simple , Follow the Huanfeng gene tutorial , Cast a successful you , Remember to pay attention to us , If you find it difficult to make a letter , Come and contact Huanfeng gene ！！！

References:

1. Li XF, Fu WF, Zhang J, Song CG. An iron metabolism and immune related gene signature for the prediction of clinical outcome and molecular characteristics of triple-negative breast cancer. BMC Cancer. 2022;22(1):619. Published 2022 Jun 7. doi:10.1186/s12885-022-09679-x