10x Genomics Support/Space Ranger 2.1/Advanced/

Space Ranger HDF5 Feature-Barcode Matrix Format

In addition to the MEX format, 10x Genomics also provides matrices in the Hierarchical Data Format (HDF5 or H5). H5 is a binary format that compresses and accesses data more efficiently than text formats such as MEX, which is useful when dealing with large datasets. H5 files are supported in both Python and R.

For more information on the format, see the Introduction to HDF5.

Note: Space Ranger generates an output file with per-molecule information in HDF5 format. Refer to the Molecule Info (H5) documentation for details.

filtered_feature_bc_matrix.h5 └── matrix [HDF5 group] ├── barcodes ├── data ├── features [HDF5 group] │ ├── _all_tag_keys │ ├── feature_type │ ├── genome │ ├── id │ ├── name │ └── target_sets [Present when --probe-set or --target-panel is specified] │ └── [target set name] ├── indices ├── indptr └── shape

The hierarchy is same for raw_feature_bc_matrix.h5 file.

The contents of the .h5 file can be examined using HDFView software or the h5dump command.

  • View all the file contents in .h5

    h5dump -n filtered_feature_bc_matrix.h5ls HDF5 "./filtered_feature_bc_matrix.h5" { FILE_CONTENTS { group / group /matrix dataset /matrix/barcodes dataset /matrix/data group /matrix/features dataset /matrix/features/_all_tag_keys dataset /matrix/features/feature_type dataset /matrix/features/genome dataset /matrix/features/id dataset /matrix/features/name group /matrix/features/target_sets dataset /matrix/features/target_sets/[target set name] dataset /matrix/indices dataset /matrix/indptr dataset /matrix/shape } }
  • View the top few lines of a specific part of .h5 (e.g. the /matrix/barcodes dataset)

    h5dump -d matrix/barcodes filtered_feature_bc_matrix.h5 | head -n 15 HDF5 "filtered_feature_bc_matrix.h5" { DATASET "matrix/barcodes" { DATATYPE H5T_STRING { STRSIZE 18; STRPAD H5T_STR_NULLPAD; CSET H5T_CSET_ASCII; CTYPE H5T_C_S1; } DATASPACE SIMPLE { ( 2965 ) / ( 2965 ) } DATA { (0): "AACACTTGGCAAGGAA-1", "AACAGGATTCATAGTT-1", "AACAGGTTATTGCACC-1", (3): "AACAGGTTCACCGAAG-1", "AACAGTCAGGCTCCGC-1", "AACAGTCCACGCGGTG-1", (6): "AACATAGTCTATCTAC-1", "AACATCTTAAGGCTCA-1", "AACCAATCTGGTTGGC-1", (9): "AACCACTGCCATAGCC-1", "AACCAGAATCAGACGT-1", "AACCGCCAGACTACTT-1", (12): "AACCGTGCTTATGTTG-1", "AACCTAAGATACTGAG-1", "AACCTACTGTAACTCA-1",

The top level of the file contains a single HDF5 group, called matrix, and metadata stored as HDF5 attributes. Within the matrix group are datasets containing the dimensions of the matrix, the matrix entries, as well as the features and spot-barcodes associated with the matrix rows and columns, respectively.

ColumnTypeDescription
barcodesstringBarcode sequences and their corresponding library identifiers (for example, AACACTTGGCAAGGAA-1). The library identifier is always -1 for spaceranger count runs of individual capture areas, and a small integer that identifies distinct capture areas in the output of spaceranger aggr
datauint32Nonzero UMI counts in column-major order
indicesuint32Zero-based row index of corresponding element in data
indptruint32Zero-based index into data / indices of the start of each column, that is the data corresponding to each barcode sequence
shapeuint64Tuple of (# rows, # columns) indicating the matrix dimensions

The matrix entries are stored in Compressed Sparse Column (CSC) format. For more details on the format, see this SciPy introduction. CSC represents the matrix in column-major order, so that each barcode is represented by a contiguous chunk of data values.

The feature reference is stored as an HDF5 group called features, within the matrix group. Note that for Targeted Gene Expression samples, the features dataset in the filtered matrix H5 file will not contain non-targeted genes, and the feature indices in target_sets are updated accordingly.

There are multiple packages that allow import of HDF5 file into R as shown in the example code below (edit path to the highlighted H5 file).

# set path to the h5 file h5_path = "/opt/sample345/outs/filtered_feature_bc_matrix.h5" # Method 1 # load the Bioconductor rhdf5 package library(rhdf5) # read in the file and examine its contents filtered_hs <- H5Fopen(h5_path) h5ls(filtered_hs) # Method 2 # load package library(Seurat) # read in the file and examine its contents filtered_hs <- Read10X_h5(h5_path) head(filtered_hs, 10)

There are two ways to load the H5 matrix into Python:

Method 1: Using cellranger.matrix module

This method requires adding spaceranger/lib/python to your $PYTHONPATH. For example, if you installed Space Ranger into /opt/spaceranger-2.0.1, then you can call the following script to set your PYTHONPATH:

$ source spaceranger-2.0.1/sourceme.bash

Then in Python, the matrix can be loaded using the cellranger.matrix module as follows (edit the path to the H5 file):

import cellranger.matrix as cr_matrix filtered_h5 = "/opt/sample345/outs/filtered_feature_bc_matrix.h5" filtered_matrix_h5 = cr_matrix.CountMatrix.load_h5_file(filtered_hs)

Method 2: Using PyTables

This method is more involved, and requires the SciPy and PyTables libraries. Edit the path to the highlighted H5 file.

import collections import scipy.sparse as sp_sparse import tables CountMatrix = collections.namedtuple('CountMatrix', ['feature_ref', 'barcodes', 'matrix']) def get_matrix_from_h5(filename): with tables.open_file(filename, 'r') as f: mat_group = f.get_node(f.root, 'matrix') barcodes = f.get_node(mat_group, 'barcodes').read() data = getattr(mat_group, 'data').read() indices = getattr(mat_group, 'indices').read() indptr = getattr(mat_group, 'indptr').read() shape = getattr(mat_group, 'shape').read() matrix = sp_sparse.csc_matrix((data, indices, indptr), shape=shape) feature_ref = {} feature_group = f.get_node(mat_group, 'features') feature_ids = getattr(feature_group, 'id').read() feature_names = getattr(feature_group, 'name').read() feature_types = getattr(feature_group, 'feature_type').read() feature_ref['id'] = feature_ids feature_ref['name'] = feature_names feature_ref['feature_type'] = feature_types tag_keys = getattr(feature_group, '_all_tag_keys').read() for key in tag_keys: feature_ref[key] = getattr(feature_group, key.decode()).read() return CountMatrix(feature_ref, barcodes, matrix) filtered_h5 = "/opt/sample345/outs/filtered_feature_bc_matrix.h5" filtered_matrix_h5 = get_matrix_from_h5(filtered_hs)

Refer to the SciPy Documentation for help.