diff --git a/README.md b/README.md
index 8c85cd9..d96381a 100644
--- a/README.md
+++ b/README.md
@@ -48,7 +48,7 @@ The whole pipeline was established based on processing graph principle. Users ca
 
 The popular WGBS/WGS analysis softwares are released on Unix/Linux system, based on different program language, like FASTQC and Bowtie2. Therefore, it's very difficult to rewrite all the software in one language. Fortunately, [conda](https://docs.conda.io/en/latest/)/[bioconda](http://bioconda.github.io/) program collected many prevalent python mudules and bioinformatics software, so we can install all the dependencies through [conda](https://docs.conda.io/en/latest/)/[bioconda](http://bioconda.github.io/) and arrange pipelines using python.
 
-We recommend using [conda/Anaconda](https://www.anaconda.com/) and create an virtual environment to manage all the dependencies. If you did not install conda before, please follow [this tutorial](https://docs.conda.io/projects/conda/en/latest/user-guide/install/) to install conda first.
+We recommend using [conda/Anaconda](https://www.anaconda.com/) and create an virtual environment to manage all the dependencies. If you did not install conda before, please follow [this tutorial](https://github.com/XWangLabTHU/cfDNApipe/blob/master/docs/conda_installation.md) to install conda first.
 
 After installation, you can create a new virtual environment for cfDNA analysis. Virtual environment management means that you can install all the dependencies in this virtual environment and delete them easily by removing this virtual environment.