Fragment assignment in the cloud with eXpress-D

BMC Bioinformatics

Table 1 Protocol Buffer specifications

Field	Type	Description
Fragment
name	string	Unique query name of fragment in SAM file
paired	bool	Boolean specifying if both ends were sequenced
alignments	FragmentAlignments	Collection of alignments for fragment
FragmentAlignment
target_id	uint32	ID of target aligned to (index in SAM header)
read_l	ReadAlignment	Alignment information for 5’ (left) read, if exists
read_r	ReadAlignment	Alignment information for 3’ (right) read, if exists
ReadAlignment
first	bool	Boolean specifying if this end was sequenced first
left_pos	unit32	0-based left endpoint of alignment to reference
right_pos	unit32	0-based right endpoint of alignment to reference
mismatch_indices	byteArray	Positions in read that differ from reference
mismatch_nucs	byteArray	Nucleotides in read at mismatches, 2 bits/nuc
Target
name	string	Unique name of target sequence
id	uint32	Index of target in SAM header
length	uint32	Number of nucleotides in target sequence
seq	byteArray	Nucleotides of target sequence, 2 bits/nuc

eXpress pre-processes the input data (SAM/BAM and FASTA file) and converts it to a format that is compatible with the distributed file system’s partitioning scheme. The information for each target and fragment are put into a space-efficient Protocol Buffer-retaining only the information necessary for optimization-, which is then serialized and encoded in base64. Each target or fragment takes up exactly one line in the file created for input into eXpress-D.

ISSN: 1471-2105