Package Bio :: Package SearchIO :: Package BlastIO
[hide private]
[frames] | no frames]

Package BlastIO

source code

Bio.SearchIO support for BLAST+ output formats.

This module adds support for parsing BLAST+ outputs. BLAST+ is a rewrite of NCBI's legacy BLAST (Basic Local Alignment Search Tool), based on the NCBI C++ toolkit. The BLAST+ suite is available as command line programs or on NCBI's web page.

Bio.SearchIO.BlastIO was tested on the following BLAST+ flavors and versions:

You should also be able to parse outputs from a local BLAST+ search or from NCBI's web interface. Although the module was not tested against all BLAST+, it should still be able to parse these other versions' outputs. Please submit a bug report if you stumble upon an unparseable file.

Some output formats from the BLAST legacy suite (BLAST+'s predecessor) may still be parsed by this module. However, results are not guaranteed. You may try to use the Bio.Blast module to parse them instead.

More information about BLAST are available through these links:

Supported Formats

Bio.SearchIO.BlastIO supports the following BLAST+ output formats:


The blast-xml parser follows the BLAST XML DTD written here:

It provides the following attributes for each SearchIO object:

Object Attribute XML Element
QueryResult target BlastOutput_db
program BlastOutput_program
reference BlastOutput_reference
version BlastOutput_version[1]
description Iteration_query-def
id Iteration_query-ID
seq_len Iteration_query-len
param_evalue_threshold Parameters_expect
param_entrez_query Parameters_entrez-query
param_filter Parameters_filter
param_gap_extend Parameters_gap-extend
param_gap_open Parameters_gap-open
param_include Parameters_include
param_matrix Parameters_matrix
param_pattern Parameters_pattern
param_score_match Parameters_sc-match
param_score_mismatch Parameters_sc-mismatch
stat_db_num Statistics_db-num
stat_db_len Statistics_db-len
stat_eff_space Statistics_eff-space
stat_entropy Statistics_entropy
stat_hsp_len Statistics_hsp-len
stat_kappa Statistics_kappa
stat_lambda Statistics_lambda
Hit accession Hit_accession
description Hit_def
id Hit_id
seq_len Hit_len
HSP bitscore Hsp_bit-score
density Hsp_density
evalue Hsp_evalue
gap_num Hsp_gaps
ident_num Hsp_identity
pos_num Hsp_positive
bitscore_raw Hsp_score
HSPFragment (also via HSP) aln_span Hsp_align-len
hit_frame Hsp_hit-frame
hit_start Hsp_hit-from
hit_end Hsp_hit-to
hit Hsp_hseq
aln_annotation Hsp_midline
pattern_start Hsp_pattern-from
pattern_end Hsp_pattern-to
query_frame Hsp_query-frame
query_start Hsp_query-from
query_end Hsp_query-to
query Hsp_qseq

You may notice that in BLAST XML files, sometimes BLAST replaces your true sequence ID with its own generated ID. For example, the query IDs become 'Query_1', 'Query_2', and so on. While the hit IDs sometimes become 'gnl|BL_ORD_ID|1', 'gnl|BL_ORD_ID|2', and so on. In these cases, BLAST lumps the true sequence IDs together with their descriptions.

The blast-xml parser is aware of these modifications and will attempt to extract the true sequence IDs out of the descriptions. So when accessing QueryResult or Hit objects, you will use the non-BLAST-generated IDs.

This behavior on the query IDs can be disabled using the 'use_raw_query_ids' parameter while the behavior on the hit IDs can be disabled using the 'use_raw_hit_ids' parameter. Both are boolean values that can be supplied to or SearchIO.parse, with the default values set to 'False'.

In any case, the raw BLAST IDs can always be accessed using the query or hit object's 'blast_id' attribute.

The blast-xml write function also accepts 'use_raw_query_ids' and 'use_raw_hit_ids' parameters. However, note that the default values for the writer are set to 'True'. This is because the writer is meant to mimic native BLAST result as much as possible.


The default format for blast-tab support is the variant without comments (-m 6 flag). Commented BLAST tabular files may be parsed, indexed, or written using the keyword argument 'comments' set to True:

# blast-tab defaults to parsing uncommented files >>> from Bio import SearchIO >>> uncommented = 'Blast/tab_2226_tblastn_004.txt' >>> qresult =, 'blast-tab') >>> qresult QueryResult(id='gi|11464971:4-101', 5 hits)

# set the keyword argument to parse commented files >>> commented = 'Blast/tab_2226_tblastn_008.txt' >>> qresult =, 'blast-tab', comments=True) >>> qresult QueryResult(id='gi|11464971:4-101', 5 hits)

For uncommented files, the parser defaults to using BLAST's default column ordering: 'qseqid sseqid pident length mismatch gapopen qstart qend sstart send evalue bitscore'.

If you want to parse an uncommented file with a customized column order, you can use the 'fields' keyword argument to pass the custom column order. The names of the column follow BLAST's naming. For example, 'qseqid' is the column for the query sequence ID. These names may be passed either as a Python list or as a space-separated strings.

# pass the custom column names as a Python list >>> fname = 'Blast/tab_2226_tblastn_009.txt' >>> custom_fields = ['qseqid', 'sseqid'] >>> qresult = next(SearchIO.parse(fname, 'blast-tab', fields=custom_fields)) >>> qresult QueryResult(id='gi|16080617|ref|NP_391444.1|', 3 hits)

# pass the custom column names as a space-separated string >>> fname = 'Blast/tab_2226_tblastn_009.txt' >>> custom_fields = 'qseqid sseqid' >>> qresult = next(SearchIO.parse(fname, 'blast-tab', fields=custom_fields)) >>> qresult QueryResult(id='gi|16080617|ref|NP_391444.1|', 3 hits)

You may also use the 'std' field name as an alias to BLAST's default 12 columns, just like when you run a command line BLAST search.

Note that the 'fields' keyword argument will be ignored if the parsed file is commented. Commented files have their column ordering stated explicitly in the file, so there is no need to specify it again in SearchIO.

'comments' and 'fields' keyword arguments are both applicable for parsing, indexing, and writing.

blast-tab provides the following attributes for each SearchIO objects:

Object Attribute Column name
QueryResult accession qacc
accession_version qaccver
gi qgi
seq_len qlen
id qseqid
Hit accession sacc
accession_version sacc_ver
gi sgi
gi_all sallgi
id_all sallseqid
seq_len slen
id sseqid
HSP bitscore bitscore
btop btop
evalue evalue
gapopen_num gapopen
gap_num gaps
ident_pct nident
ident_num pident
mismatch_num mismatch
pos_pct ppos
pos_num positive
bitscore_raw score
HSPFragment (also via HSP) frames frames[2]
aln_span length
query_end qend
query_frame qframe
query qseq
query_start qstart
hit_end send
hit_frame sframe
hit sseq
hit_start sstart

If the parsed file is commented, the following attributes may be available as well:

Object Attribute Value
QueryResult description query description
fields columns in the output file
program BLAST flavor
rid remote search ID
target target database
version BLAST version


The BLAST plain text output format has been known to change considerably between BLAST versions. NCBI itself has recommended that users not rely on the plain text output for parsing-related work.

However, in some cases parsing the plain text output may still be useful. SearchIO provides parsing support for the plain text output, but guarantees only a minimum level of support. Writing a parser that fully supports plain text output for all BLAST versions is not a priority at the moment.

If you do have a BLAST plain text file that can not be parsed and would like to submit a patch, we are more than happy to accept it.

The blast-text parser provides the following object attributes:

Object Attribute Value
QueryResult description query sequence description
id query sequence ID
program BLAST flavor
seq_len full length of query sequence
target target database of the search
version BLAST version
Hit evalue hit-level evalue, from the hit table
id hit sequence ID
description hit sequence description
score hit-level score, from the hit table
seq_len full length of hit sequence
HSP evalue hsp-level evalue
bitscore hsp-level bit score
bitscore_raw hsp-level score
gap_num number of gaps in alignment
ident_num number of identical residues in alignment
pos_num number of positive matches in alignment
HSPFragment (also via HSP) aln_annotation alignment similarity string
aln_span length of alignment fragment
hit hit sequence
hit_end hit sequence end coordinate
hit_frame hit sequence reading frame
hit_start hit sequence start coordinate
hit_strand hit sequence strand
query query sequence
query_end query sequence end coordinate
query_frame query sequence reading frame
query_start query sequence start coordinate
query_strand query sequence strand
[1]may be modified
[2]When 'frames' is present, both query_frame and hit_frame will be present as well. It is recommended that you use these instead of 'frames' directly.
Submodules [hide private]

Variables [hide private]
  __package__ = 'Bio.SearchIO.BlastIO'