{"id":23685,"date":"2024-01-08T17:27:35","date_gmt":"2024-01-08T09:27:35","guid":{"rendered":"http:\/\/www.biocloudservice.com\/wordpress\/?p=23685"},"modified":"2024-01-08T17:27:36","modified_gmt":"2024-01-08T09:27:36","slug":"%e7%9b%b4%e5%87%bb%e7%83%ad%e7%82%b9%ef%bc%81%e7%94%9f%e4%bf%a1%e6%96%87%e7%ab%a0%e5%bf%85%e5%a4%87%e4%b9%8blassorandomforest%e5%92%8cxgboost%e4%b8%89%e7%a7%8d%e6%9c%ba%e5%99%a8%e5%ad%a6%e4%b9%a0","status":"publish","type":"post","link":"http:\/\/www.biocloudservice.com\/wordpress\/?p=23685","title":{"rendered":"\u76f4\u51fb\u70ed\u70b9\uff01\u751f\u4fe1\u6587\u7ae0\u5fc5\u5907\u4e4bLASSO,randomForest\u548cxgboost\u4e09\u79cd\u673a\u5668\u5b66\u4e60\u7b97\u6cd5\u8fdb\u884c\u7279\u5f81\u57fa\u56e0\u7b5b\u9009"},"content":{"rendered":"

\u5c0f\u679c\u4eca\u5929\u7ed9\u5927\u5bb6\u5e26\u6765\u7684\u5206\u4eab\u7edd\u5bf9\u662f\u751f\u4fe1\u70ed\u70b9\u5206\u6790\u5185\u5bb9\uff0c\u57fa\u4e8e\u4e09\u79cd\u673a\u5668\u5b66\u4e60\u7b97\u6cd5\u8fdb\u884c\u751f\u5b58\u8d44\u6599\u7279\u5f81\u57fa\u56e0\u7b5b\u9009\uff0c\u6700\u7ec8\u5c06\u4e09\u79cd\u7b97\u6cd5\u83b7\u5f97\u7684\u4ea4\u53c9\u57fa\u56e0\u4f5c\u4e3a\u6211\u4eec\u7684\u7279\u5f81\u57fa\u56e0\uff0c\u63a5\u4e0b\u6765\u8ddf\u7740\u5c0f\u679c\u5f00\u542f\u4eca\u5929\u7684\u5b66\u4e60\u4e4b\u65c5\u5427\uff01<\/p>\n

    \n
  1. \u4e09\u79cd\u673a\u5668\u5b66\u4e60\u7b97\u6cd5\u8fdb\u884c\u7279\u5f81\u7b5b\u9009<\/li>\n<\/ol>\n

    \u5c0f\u4f19\u4f34\u4eec\u662f\u4e0d\u662f\u5728\u60f3\u5230\u5e95\u662f\u90a3\u4e09\u79cd\u673a\u5668\u5b66\u4e60\u7b97\u6cd5\uff1f\u4e0d\u6025\uff01\u5c0f\u679c\u6765\u6162\u6162\u544a\u8bc9\u5927\u5bb6\uff0c\u4eca\u5929\u5c0f\u679c\u5c06\u5229\u7528LASSO\u56de\u5f52\uff0c\u968f\u673a\u68ee\u6797\u548cxgboost\u4e09\u79cd\u7b97\u6cd5\u8fdb\u884c\u751f\u5b58\u8d44\u6599\u7279\u5f81\u57fa\u56e0\u7b5b\u9009\uff0c\u8fd9\u4e09\u79cd\u673a\u5668\u5b66\u4e60\u7684\u7b97\u6cd5\u539f\u7406\u5c0f\u679c\u5728\u8fd9\u91cc\u4e0d\u505a\u8fc7\u591a\u7684\u4ecb\u7ecd\uff0c\u611f\u5174\u8da3\u7684\u5c0f\u4f19\u4f34\u4eec\u53ef\u4ee5\u81ea\u884c\u67e5\u8be2\u5b66\u4e60\uff0c\u4eca\u5929\u5c0f\u679c\u5c06\u901a\u8fc7\u5b9e\u64cd\uff0c\u8ba9\u5927\u5bb6\u8f7b\u677e\u638c\u63e1\u5982\u4f55\u5229\u7528\u8fd9\u4e09\u79cd\u7b97\u6cd5\u8fdb\u884c\u7279\u5f81\u57fa\u56e0\u7b5b\u9009\uff0c\u5176\u5b9e\u5206\u6790\u975e\u5e38\u7b80\u5355\uff0c\u53ea\u9700\u8981\u8f93\u5165\u5e26\u6709\u751f\u5b58\u4fe1\u606f\u7684\u8868\u8fbe\u77e9\u9635\u6587\u4ef6\uff0c\u5c31\u53ef\u4ee5\u5f88\u8f7b\u677e\u7684\u5b8c\u6210\u8be5\u5206\u6790\uff0c\u5bf9\u5c0f\u767d\u4e5f\u5f88\u9002\u7528\uff0c\u8bdd\u4e0d\u591a\u8bf4\uff0c\u9a6c\u4e0a\u8ddf\u7740\u5c0f\u679c\u5f00\u59cb\u4eca\u5929\u7684\u5b9e\u64cd\u5427\uff01<\/p>\n

      \n
    1. \u51c6\u5907\u9700\u8981\u7684R\u5305<\/li>\n<\/ol>\n

      #\u5b89\u88c5\u9700\u8981\u7684R\u5305<\/p>\n

      install.packages(“ggplot2”)<\/p>\n

      install.packages(“tidyverse”)<\/p>\n

      install.packages(“ggvenn”)<\/p>\n

      install.packages(“glmnet”)<\/p>\n

      install.packages(“survival”)<\/p>\n

      install.packages(“xgboost”)<\/p>\n

      install.packages(“randomForestSRC”)<\/p>\n

      #\u52a0\u8f7d\u9700\u8981\u7684R\u5305<\/p>\n

      library(randomForestSRC)<\/p>\n

      library(tidyverse)<\/p>\n

      library(ggplot2)<\/p>\n

      library(ggvenn)<\/p>\n

      library(glmnet)<\/p>\n

      library(survival)<\/p>\n

      library(xgboost)<\/p>\n

        \n
      1. \u6570\u636e\u51c6\u5907<\/li>\n<\/ol>\n

        exp_surv.txt<\/p>\n

        #\u751f\u5b58\u4fe1\u606f\u52a0\u57fa\u56e0\u8868\u8fbe\u77e9\u9635\u6587\u4ef6\uff0c\u884c\u540d\u4e3a\u6837\u672c\u4fe1\u606f\uff0c\u7b2c\u4e00\u5217\u4e3aOS.time(\u751f\u5b58\u65f6\u95f4),\u7b2c\u4e8c\u5217\u4e3aOS(\u751f\u5b58\u72b6\u6001),\u5176\u4ed6\u5217\u4e3a\u57fa\u56e0\u3002<\/p>\n

        \"Dingtalk_20230505091052\"<\/p>\n

        4 . LASSO\u56de\u5f52\u5206\u6790<\/p>\n

        #\u8bfb\u53d6\u751f\u5b58\u4fe1\u606f\u7684\u57fa\u56e0\u8868\u8fbe\u77e9\u9635<\/p>\n

        train<-read.table(“exp_surv.txt”,header=T,sep=”\\t”)<\/p>\n

        #\u63d0\u53d6\u57fa\u56e0\u8868\u8fbe\u77e9\u9635<\/p>\n

        x1 <- as.matrix(train[, 3:ncol(train)])<\/p>\n

        x2 <- as.matrix(Surv(train$OS.time, train$OS))<\/p>\n

        cvfit = cv.glmnet(x1, x2,<\/p>\n

        nfold=10,<\/p>\n

        family = “cox”<\/p>\n

        )<\/p>\n

        #\u7ed8\u5236lamba\u56fe<\/p>\n

        pdf(“lambda.pdf”,width = 10,height = 8)<\/p>\n

        plot(cvfit)<\/p>\n

        dev.off()<\/p>\n

        #\u6b64\u5904\u4f7f\u7528`lambda min`<\/p>\n

        cvfit$lambda.min #\u6700\u4f73lambda\u503c<\/p>\n

        \"Dingtalk_20230505135037\"<\/p>\n

        fit <- glmnet(x1, x2,<\/p>\n

        family = “cox”)<\/p>\n

        #\u7528\u5305\u81ea\u5e26\u7684\u51fd\u6570\u7ed8\u5236lasso\u56de\u5f52\u56fe<\/p>\n

        pdf(“lasso.pdf”,width = 10,height = 8)<\/p>\n

        plot(fit)<\/p>\n

        dev.off()<\/p>\n

        #\u6700\u5c0f\u503c\u4e00\u4e2a\u6807\u51c6\u8bef\u65f6\u5404\u53d8\u91cf\u7684\u7cfb\u6570\u503c<\/p>\n

        coef.min = coef(cvfit, s = “lambda.min”)<\/p>\n

        #\u975e\u96f6\u7cfb\u6570<\/p>\n

        active.min = which(coef.min != 0)<\/p>\n

        #\u975e\u96f6\u53d8\u91cf\uff0c\u57fa\u56e0\u540d<\/p>\n

        geneids <- rownames(coef.min)[active.min]<\/p>\n

        #\u63d0\u53d6\u9009\u4e2d\u7684\u57fa\u56e0\u5bf9\u5e94\u7684coefficient\uff0c\u56de\u5f52\u7cfb\u6570<\/p>\n

        index.min = coef.min[active.min]<\/p>\n

        #\u6309\u5217\u5408\u5e76\u4e24\u4e2a\u53d8\u91cf<\/p>\n

        combine <- cbind(geneids, index.min)<\/p>\n

        #\u5c06\u6570\u636e\u8f6c\u6362\u4e3a\u6570\u636e\u6846<\/p>\n

        lasso.result<-as.data.frame(combine)<\/p>\n

        #\u5c06index.min\u6570\u636e\u7c7b\u578b\u6539\u4e3a\u6570\u503c\u578b\u3002<\/p>\n

        lasso.result$index.min<-as.numeric(lasso.result$index.min)<\/p>\n

        \"Dingtalk_20230505135150\"<\/p>\n

        #\u901a\u8fc7LASSO\u56de\u5f52\u7b5b\u9009\u7684\u7279\u5f81\u57fa\u56e0\u6587\u4ef6\uff0c\u7b2c\u4e00\u5217\u8868\u793a\u57fa\u56e0\u540d\uff0c\u7b2c\u4e8c\u5217\u8868\u793acoefficient\u3002<\/p>\n

        5.\u968f\u673a\u68ee\u6797\u5206\u6790<\/p>\n

        fit <- rfsrc(Surv(OS.time,OS)~., data = train,<\/p>\n

        ntree = 1000, nodesize = 14,<\/p>\n

        splitrule = ‘logrank’,<\/p>\n

        importance = T,<\/p>\n

        proximity = T,<\/p>\n

        forest = T)<\/p>\n

        #\u7ed8\u5236OBB\u56fe\u548cVIP\u56fe<\/p>\n

        pdf(“forest.pdf”)<\/p>\n

        plot(fit)<\/p>\n

        dev.off()<\/p>\n

        #\u7b5b\u9009\u7279\u5f81\u57fa\u56e0<\/p>\n

        rftop<-var.select(fit)<\/p>\n

        #\u5c06\u6570\u636e\u7c7b\u578b\u8f6c\u5316\u4e3a\u6570\u636e\u6846<\/p>\n

        rftop2<-data.frame(<\/p>\n

        Feature=rftop$topvars,<\/p>\n

        vimp=rftop$varselect[rftop$topvars,2])<\/p>\n

        \"Dingtalk_20230505135426\"<\/p>\n

        \u6ce8\uff1a\u901a\u8fc7\u968f\u673a\u68ee\u6797\u7b97\u6cd5\u7b5b\u9009\u7684\u7279\u5f81\u57fa\u56e0\uff0c\u7b2c\u4e00\u5217\u8868\u793a\u57fa\u56e0\u540d\uff0c\u7b2c\u4e8c\u5217\u4e3aVIP\u503c\u3002<\/p>\n

        6.xgboost\u7b97\u6cd5<\/p>\n

        x_label<-ifelse(train$OS==1,train$OS.time,-train$OS.time)<\/p>\n

        x_train<-as.matrix(train[,c(3:ncol(train))])<\/p>\n

        x_val<-xgb.DMatrix(x_train,<\/p>\n

        label=x_label)<\/p>\n

        xgb_param<-list(“objective”=”survival:cox”,<\/p>\n

        “eval_metric”=”cox-nloglik”)<\/p>\n

        xgb.model<-xgb.train(params=xgb_param,<\/p>\n

        data=x_val,<\/p>\n

        nrounds=100,<\/p>\n

        eta=0.1,<\/p>\n

        watchlist=list(val2=x_val))<\/p>\n

        e<-data.frame(xgb.model$evaluation_log)<\/p>\n

        ggplot(aes(x=iter,y=val2_cox_nloglik),data=e)+<\/p>\n

        geom_point(color=”pink”,size=3)+<\/p>\n

        geom_line()+<\/p>\n

        theme_bw()<\/p>\n

        ggsave(“train_error.pdf”)<\/p>\n

        #\u7b5b\u9009VIP\u503c\u6392\u540d\u524d\u5341\u7684\u57fa\u56e0\u3002<\/p>\n

        imp<-xgb.importance(feature_names=colnames(x_train),model=xgb.model)<\/p>\n

        top_10<-imp[1:10,1:2]<\/p>\n

        \"Dingtalk_20230505135748\"<\/p>\n

        \u6ce8\uff1a\u901a\u8fc7xgboost\u7b97\u6cd5\u7b5b\u9009\u7684VIP\u503c\u6392\u524d\u5341\u7684\u57fa\u56e0\uff0c\u7b2c\u4e00\u5217\u8868\u793a\u57fa\u56e0\u540d\uff0c\u7b2c\u4e8c\u5217\u8868\u793aVIP\u503c\u3002<\/p>\n

          \n
        1. \u7ed8\u5236venn\u56fe\u7b5b\u9009\u4e09\u79cd\u7b97\u6cd5\u83b7\u5f97\u7684\u4ea4\u53c9\u57fa\u56e0<\/li>\n<\/ol>\n

          gene<-list(LASSO=geneids,<\/p>\n

          RF=rftop2$Feature\uff0c<\/p>\n

          Xgboost=top_10$Feature)<\/p>\n

          ggvenn(gene,<\/p>\n

          fill_color=c(“#0073C2FF”,”#EFC000FF”,\u201d#CD534CFF\u201d),<\/p>\n

          stroke_size=0.5,<\/p>\n

          set_name_size=4)<\/p>\n

          ggsave(“venn.pdf”)<\/p>\n

          7.\u7ed3\u679c\u6587\u4ef6<\/p>\n

            \n
          1. lambda.pdf<\/li>\n<\/ol>\n

            \u8be5\u7ed3\u679c\u56fe\u7247\u4e3aLASSO\u56de\u5f52lambda\u56fe\u3002<\/p>\n

            \"Dingtalk_20230505094959\"<\/p>\n

              \n
            1. lasso.pdf<\/li>\n<\/ol>\n

              \u8be5\u7ed3\u679c\u56fe\u7247LASSO\u56de\u5f52\u4e3a10\u6298\u4ea4\u53c9\u9a8c\u8bc1\u56fe\u3002<\/p>\n

              \"Dingtalk_20230505102237\"<\/p>\n

                \n
              1. forest.pdf<\/li>\n<\/ol>\n

                \u8be5\u7ed3\u679c\u56fe\u7247\u4e3a\u968f\u673a\u68ee\u6797OBB\u56fe\u548cVIP\u56fe\u3002<\/p>\n

                \"Dingtalk_20230505102421\"<\/p>\n

                  \n
                1. train_error.pdf<\/li>\n<\/ol>\n

                  \u8be5\u7ed3\u679c\u56fe\u7247\u4e3axgboost\u7b97\u6cd5cox\u98ce\u9669\u6bd4\u4f8b\u56de\u5f52\u7684\u8d1f\u504f\u5bf9\u6570\u968f\u7740\u8fed\u4ee3\u6b21\u6570\u7684\u53d8\u5316\u3002<\/p>\n

                  \"Dingtalk_20230505102954\"<\/p>\n

                    \n
                  1. venn.pdf<\/li>\n<\/ol>\n

                    \u8be5\u7ed3\u679c\u56fe\u7247\u4e3a\u4e09\u79cd\u7b97\u6cd5\u7279\u5f81\u57fa\u56e0\u7684venn\u56fe\uff0c\u5c06\u4ea4\u53c9\u57fa\u56e0\u7b5b\u9009\u4e3a\u5019\u9009\u57fa\u56e0\u3002<\/p>\n

                    \"Dingtalk_20230505103538\"<\/p>\n

                    \u6700\u7ec8\u5c0f\u679c\u987a\u5229\u5b8c\u6210\u4e86\u5229\u7528 lassso\u56de\u5f52,\u968f\u673a\u68ee\u6797\u548cxgboost\u4e09\u79cd\u673a\u5668\u5b66\u4e60\u7b97\u6cd5\u8fdb\u884c\u4e86\u7279\u5f81\u57fa\u56e0\u7b5b\u9009\uff0c\u673a\u5668\u5b66\u4e60\u76f8\u5173\u5176\u4ed6\u5206\u6790\u5185\u5bb9\u6b22\u8fce\u5c1d\u8bd5\u672c\u516c\u53f8\u65b0\u5f00\u53d1\u7684\u4e91\u5e73\u53f0\u751f\u7269\u4fe1\u606f\u5206\u6790\u5c0f\u5de5\u5177\uff0c\u96f6\u4ee3\u7801\u5b8c\u6210\u5206\u6790\uff0c\u4e91\u5e73\u53f0\u7f51\u5740\uff1ahttp:\/\/www.biocloudservice.com\/home.html\uff0c\u4e3b\u8981\u5305\u62eclasso\u56de\u5f52\u6a21\u578b\u7b5b\u9009\u7279\u5f81\u57fa\u56e0(http:\/\/www.biocloudservice.com\/116\/116.php),\u968f\u673a\u68ee\u6797\u7684\u5341\u6298\u4ea4\u53c9\u9a8c\u8bc1<\/p>\n

                    (http:\/\/www.biocloudservice.com\/646\/646.php),svm-ref\u652f\u6301\u5411\u91cf\u673a\u7684\u673a\u5668\u5b66\u4e60(http:\/\/www.biocloudservice.com\/372\/372.php)\u7b49\u673a\u5668\u5b66\u4e60\u76f8\u5173\u7684\u5c0f\u5de5\u5177\uff0c\u6b22\u8fce\u5927\u5bb6\u6765\u5c1d\u8bd5\u54df\uff01<\/p>\n","protected":false},"excerpt":{"rendered":"

                    \u5c0f\u679c\u4eca\u5929\u7ed9\u5927\u5bb6\u5e26\u6765\u7684\u5206\u4eab\u7edd\u5bf9\u662f\u751f\u4fe1\u70ed\u70b9\u5206\u6790\u5185\u5bb9\uff0c\u57fa\u4e8e\u4e09\u79cd\u673a\u5668\u5b66\u4e60\u7b97\u6cd5\u8fdb\u884c\u751f\u5b58\u8d44\u6599\u7279\u5f81\u57fa\u56e0\u7b5b\u9009\uff0c\u6700\u7ec8\u5c06\u4e09\u79cd\u7b97\u6cd5\u83b7 […]<\/p>\n","protected":false},"author":1,"featured_media":0,"comment_status":"closed","ping_status":"closed","sticky":false,"template":"","format":"standard","meta":{"jetpack_post_was_ever_published":false,"_jetpack_newsletter_access":""},"categories":[1],"tags":[],"jetpack_featured_media_url":"","_links":{"self":[{"href":"http:\/\/www.biocloudservice.com\/wordpress\/index.php?rest_route=\/wp\/v2\/posts\/23685"}],"collection":[{"href":"http:\/\/www.biocloudservice.com\/wordpress\/index.php?rest_route=\/wp\/v2\/posts"}],"about":[{"href":"http:\/\/www.biocloudservice.com\/wordpress\/index.php?rest_route=\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"http:\/\/www.biocloudservice.com\/wordpress\/index.php?rest_route=\/wp\/v2\/users\/1"}],"replies":[{"embeddable":true,"href":"http:\/\/www.biocloudservice.com\/wordpress\/index.php?rest_route=%2Fwp%2Fv2%2Fcomments&post=23685"}],"version-history":[{"count":1,"href":"http:\/\/www.biocloudservice.com\/wordpress\/index.php?rest_route=\/wp\/v2\/posts\/23685\/revisions"}],"predecessor-version":[{"id":23696,"href":"http:\/\/www.biocloudservice.com\/wordpress\/index.php?rest_route=\/wp\/v2\/posts\/23685\/revisions\/23696"}],"wp:attachment":[{"href":"http:\/\/www.biocloudservice.com\/wordpress\/index.php?rest_route=%2Fwp%2Fv2%2Fmedia&parent=23685"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"http:\/\/www.biocloudservice.com\/wordpress\/index.php?rest_route=%2Fwp%2Fv2%2Fcategories&post=23685"},{"taxonomy":"post_tag","embeddable":true,"href":"http:\/\/www.biocloudservice.com\/wordpress\/index.php?rest_route=%2Fwp%2Fv2%2Ftags&post=23685"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}