Linux 下用R语言打开hdf5(.h5)文件异常的解决方案

6月10日从GEO数据库下载了作者上传的表达矩阵h5格式文件，需用从中提取出表达矩阵然后重构Seurat分析对象；该exp_matrix.hdf5文件在我的Window系统上，可以通过rhdf5::H5Fopen和hdf5r::H5File$new两种方法打开读取文件内容。但是由于文件太大，有6个G，在我24G内存的Windows系统上没办法完全加载出矩阵。于是就把文件上传到服务器集群进行处理，以同样的方式尝试打开exp_matrix.hdf5文件，碰到了以下问题：

• 1.rhdf5::H5Fopen函数打开error：

data.h5 <- H5Fopen(name = "exp_matrix.hdf5")
Error in H5Fopen(name = "exp_matrix.hdf5",  : 
  HDF5. File accessibility. Unable to open file.

• 2.hdf5r::H5File$new函数打开error：

hdf5r::H5File$new("exp_matrix.hdf5", mode="r")
Error in H5File.open(filename, mode, file_create_pl, file_access_pl) : 
  HDF5-API Errors:
    error #000: H5F.c in H5Fopen(): line 620: unable to open file
        class: HDF5
        major: File accessibility
        minor: Unable to open file

    error #001: H5VLcallback.c in H5VL_file_open(): line 3501: failed to iterate over available VOL connector plugins
        class: HDF5
        major: Virtual Object Layer
        minor: Iteration failed

    error #002: H5PLpath.c in H5PL__path_table_iterate(): line 578: can't iterate over plugins in plugin path '(null)'
        class: HDF5
        major: Plugin for dynamically loaded library
        minor: Iteration failed

    error #003: H5PLpath.c in H5PL__path_table_iterate_process_path(): line 620: can't open directory: ~/local/Minconda/envs/scRNA/lib/hdf5/plugin
        class: HDF5
        major: Plugin for dynamically loaded library
        minor: Can't open directory or file

    error #004: H5VLcallback.c in H5VL__file_open(): line 3351: open faile

解决方案：编辑Linux用户的配置文件(.bashrc)添加以下两句环境变量：

export HDF5_USE_FILE_LOCKING="FALSE"
export RHDF5_USE_FILE_LOCKING="FALSE" 

然后执行source ~/.bashrc,就可以解决Linux环境下无法读取h5文件的错误了。

Error in H5Fopen(file): Unable to open HDF5 file when using subsetArchRProject to create new arrow files · Issue #248 · GreenleafLab/ArchR (github.com)