Accepted Read Data Formats¶

Single cell read data¶

Single cell read data must be submitted in the BAM or CRAM format using the following tags specified in the SAM Optional Fields Specification:

CB: Cell identifier
CR: Cellular barcode sequence bases (uncorrected)
CY: Phred quality of the cellular barcode sequence in the CR tag

Sample de-multiplexing¶

Reads for different samples should be submitted using separate files. The only exception is when a BAM or CRAM file contains reads for a large number of samples intented to be always analysed together. In this case the sample associated with the read file should describe the sample group while the BAM or CRAM file should identify the sample for each read.

Standard formats¶

The following standard file formats are accepted and transformed into Fastq products:

cram
bam
fastq

CRAM format¶

Each submitted CRAM file must:

be compatible with the CRAM Format Specification
be readable with Samtools
contain only reference sequences that exist in the CRAM Reference Registry
be submitted as a separate run
use the .cram file name suffix (e.g. ‘a.cram’)

CRAM file names are required to end up with the .cram suffix (e.g. ‘a.cram’).

A CRAM index (CRAI) file is created by the archive for each submitted CRAM file and is available in the same directory as the CRAM file from which is was created. CRAM index file names start with the CRAM file name and end up with the .crai suffix (e.g. ‘a.cram.crai’ for CRAM file ‘a.cram’).

BAM format¶

Each submitted BAM file must:

be compatible with the SAM/BAM Format Specification
be readable with Samtools
be submitted as a separate run
use the .bam file name suffix (e.g. ‘a.bam’)

PacBio BAM files¶

We support the submission of the following types of PacBio BAM files:

subread BAM files (*.subreads.bam)
CCS read BAM files (*.ccs.bam)

Fastq format¶

We recommend that read data is either submitted in BAM or CRAM format. However, single and paired reads are accepted as Fastq files that meet the following the requirements:

Quality scores must be in Phred scale.
Both ASCII and space delimitered decimal encoding of quality scores are supported. We will automatically detect the Phred quality offset of either 33 or 64.
No technical reads (e.g. adapters, linkers, barcodes, primers) are allowed.
Single reads must be submitted using a single Fastq file and can be submitted with or without read names.
Paired reads must be submitted using two Fastq files.
The first line for each read must start with ‘@’.
The base calls and quality scores must be separated by a line starting with ‘+’.
Paired read names must either use Casava 1.8 read names (regular expression: ^@([a-zA-Z0-9_-]+:[0-9]+:[a-zA-Z0-9_-]+:[0-9]+:[0-9]+:[0-9-]+:[0-9-]+) ([12]):[YN]:[0-9]*[02468]:[ACGTN]+$) or must end with /1 or /2 optionally followed by a space and a comment.
The Fastq files must be compressed using gzip or bzip2.
The regular expression for bases is “^([ACGTNactgn.]*?)$”

Example of Fastq file containing single reads:

@read_name
GATTTGGGGTTCAAAGCAGTATCGATCAAATAGTAAATCCATTTGTTCAACTCACAGTTT
+
!''*((((***+))%%++)(%%).1***-+*''))**55CCF>>>>>>CCCCCCC65
...

Example of Fastq file containing paired reads (prior to Casava 1.8):

@read_name/1
GATTTGGGGTTCAAAGCAGTATCGATCAAATAGTAAATCCATTTGTTCAACTCACAGTTT
+
!''*((((***+))%%++)(%%).1***-+*''))**55CCF>>>>>>CCCCCCC65
@read_name/2
GATTTGGGGTTCAAAGCAGTATCGATCAAATAGTAAATCCATTTGTTCAACTCACAGTTT
+
!''*((((***+))%%++)(%%).1***-+*''))**55CCF>>>>>>CCCCCCC65
...

With Casava 1.8 the format of the ‘@’ line has changed and we accept this pattern too:

@EAS139:136:FC706VJ:2:2104:15343:197393 1:Y:18:ATCACG

Platform specific formats¶

Oxford Nanopore¶

Oxford Nanopore native data must be submitted as a single tar.gz archive containing basecalled fast5 files from Guppy, Metrichor, or Albacore.

For Metrichor, an example directory structure for run named XYZ:

XYZ/reads/downloads/fail/
XYZ/reads/downloads/pass/

How to archive all files in the XYZ downloads directory in a linux command line:

cd <directory containing XYZ directory>
tar -cvzf XYZ.tar.gz XYZ/reads/downloads/

PacBio¶

PacBio data submissions are supported in the platform specific native format.

One run consists of *.bax.h5, *.bas.h5 and xml files. Please note that these files must not be tarred.

SFF format¶

The SFF format is supported for the 454 and Ion Torrent platforms.

10x Genomics¶

To submit 10x Genomics data where read indexes exist, you must convert to BAM or CRAM format. The supported tags are defined in the SAM Optional Fields Specification

Formats being deprecated¶

Complete Genomics native format¶

The full Complete Genomics data package can be submitted including the ASM, LIB and MAP subfolders. Each data package should be submitted as a single experiment and run. Please note the data package must not be tarred or gzipped for submission.

SRF format¶

The *_seq.txt files can be converted into SRF files using the illumina2srf utility available from the DNA Sequence Read Toolkit.

Each Illumina lane should be submitted as a separate SRF file and runs should be demultiplexed prior SRF file generation.

To produce a SRF submission file for a non-paired lane, change the working directory to the run folder and run:

illumina2srf -R -P -N <run>:%l:%t: -n %x:%y -o <center_name>_<run>_<lane>.srf s_<lane>_*_seq.txt

The -R, -P options are used to exclude intensity, noise and signal data from the generated SRF files. These data series are no longer supported for new data submissions.

The recommended format for the SRF file names is <center_name>.srf, where <center_name> is the center name abbreviation assigned to all submitters, and the and are the run and the lane identifiers.

To produce a SRF submission file for paired lane, change the working directory to the run folder and run:

illumina2srf -R -P -N <run>:%l:%t: -n %x:%y -2 <cycle> -o <center_name>_<run>_<lane>.srf s_<lane>_*_seq.txt

Deprecated formats¶

Read submissions are no longer accepted in the following formats:

SOLiD csfasta/qual format (support ended in 2015)
Illumina qseq format (support ended in 2015)
Illumina scarf format (support ended in 2015)