BLAST processing rules

[Home]

[FAQ]

HSP Processing: NCBI BLAST 2.0 series and PSI-BLAST

The rules used in MView for selecting and tiling HSPs. The command line option -hsp mode is used to choose between ranked (the default), all, or discrete processing.

Differences between NCBI BLAST 1.4 and NCBI BLAST 2.0

• In the default gapped mode, the BLAST 2.0 programs produce no 'N' fragment counts. Fragment counts are switched on in BLAST when it is run in ungapped mode using the 'blastall... -g F' command line option.
• Significances are reported only as Expectation (e-values) 'E' rather than p-values 'P'.
• BLASTN: negative orientation query strands are not reported because the query and hit can always be reverse complemented to revert the query back to the positive strand orientation.
• BLASTX: query strands don't carry frame information (early versions only).
• TBLASTN: hit strands don't carry frame information (early versions only).
• TBLASTX: query and hit strands don't carry frame information (early versions only).

With BLAST 2.0 series output, MView assumes that only one fragment (a gapped alignment) is required in 'ranked' mode and accepts the FIRST alignment with score and expectation matching those in the hit ranking. If you wish to see the effect of any other HSPs use the '-hsp all' or '-hsp discrete' options to MView. This is particularly important if you use BLAST in ungapped mode.

BLASTP, PSI-BLAST
Input		Output		Orient		Assemble	Renumber	Filter
qry	hit	qry	hit	qry	hit
pro	pro	pro	pro	+	+	simple	no	none
pro	pro	pro	pro	+	+	simple	no	cycle
For each specified search cycle (PSI-BLAST only):
-hsp ranked		For each ranked hit, the first alignment is selected to form an alignment row using: (i) same expectation 'E', (ii) same score 'bits'. Row data fields: (i) rank as integral 'hitnum', (ii) identifier, (iii) description, (iv) ranked 'bits', (v) ranked 'E', (vi) 'N'=1.
-hsp all		For each ranked hit, all alignments are selected for assembly. Row data fields: (i) rank as integral 'hitnum', (ii) identifier, (iii) description, (iv) highest 'bits', (v) lowest 'E', (vi) total number of alignments 'N'.
-hsp discrete		For each ranked hit, every alignment is assigned its own row. Row data fields: (i) alignment rank as compound rank 'hitnum.alignmentnum', (ii) identifier, (iii) description, (iv) alignment 'bits', (v) alignment 'E', (vi) 'N'=1.
BLASTN
Input		Output		Orient		Assemble	Renumber	Filter
qry	hit	qry	hit	qry	hit
dna	dna	dna	dna	+	+	simple	no	none
				+	-	simple	no	none
For a given query orientation (+, -) derived from the query sequence numbering:
-hsp ranked		For each ranked hit, alignments corresponding to those in the ranking are selected for assembly having: (i) same expectation 'E', (ii) same score 'bits'. Row data fields: (i) rank as integral 'hitnum', (ii) identifier, (iii) description, (iv) ranked 'bits', (v) ranked 'E', (vi) 'N'=1. (vii) 'hit' orientation.
-hsp all		For each ranked hit, alignments are selected for assembly having: (i) same hit orientation. Row data fields: (i) rank as compound 'hitnum.alignmentnum', (ii) identifier, (iii) description, (iv) highest 'bits', (v) lowest 'E', (vi) total number of alignments 'N', (vii) 'hit' orientation.
-hsp discrete		For each ranked hit, every alignment is assigned its own row. Row data fields: (i) alignment rank as compound rank 'hitnum.alignmentnum', (ii) identifier, (iii) description, (iv) alignment 'bits', (v) alignment 'E', (vi) 'N'=1, (vii) 'hit' orientation.
BLASTX
Input		Output		Orient		Assemble	Renumber	Filter
qry	hit	qry	hit	qry	hit
dna	pro	pro	pro	+	+	simple	yes	query orient
				-	+	reverse	yes	query orient
For a given query orientation (+, -) derived from the query sequence numbering:
-hsp ranked		For each ranked hit, alignments corresponding to those in the ranking are selected for assembly having: (i) same expectation 'E', (ii) same score 'bits', (iii) same query orientation. Row data fields: (i) rank as integral 'hitnum', (ii) identifier, (iii) description, (iv) ranked 'bits', (v) ranked 'E', (vi) 'N=1', (vii) query orientation 'qry'.
-hsp all		For each ranked hit, alignments are selected for assembly having: (i) same query orientation. Row data fields: (i) rank as integral 'hitnum', (ii) identifier, (iii) description, (iv) highest 'bits', (v) lowest 'E', (vi) total number of alignments 'N', (vii) query orientation 'qry'.
-hsp discrete		For each ranked hit, every alignment is assigned its own row. Row data fields: (i) alignment rank as compound rank 'hitnum.alignmentnum', (ii) identifier, (iii) description, (iv) alignment 'bits', (v) alignment 'E', (vi) 'N'=1, (vii) query orientation 'qry'.
TBLASTN
Input		Output		Orient		Assemble	Renumber	Filter
qry	hit	qry	hit	qry	hit
pro	dna	pro	pro	+	+	simple	no	none
				+	-	simple	no	none
-hsp ranked		For each ranked hit, alignments corresponding to those in the ranking are selected for assembly having: (i) same expectation 'E', (ii) same score 'bits', (iii) same hit orientation. Row data fields: (i) rank as integral 'hitnum', (ii) identifier, (iii) description, (iv) ranked 'bits', (v) ranked 'E', 'N'=1, (vii) 'hit' orientation.
-hsp all		For each ranked hit, alignments are selected for assembly having: (i) same hit orientation. Row data fields: (i) rank as compound 'hitnum.alignmentnum', (ii) identifier, (iii) description, (iv) highest 'bits', (v) lowest 'E', (vi) total number of alignments 'N', (vii) 'hit' orientation.
-hsp discrete		For each ranked hit, every alignment is assigned its own row. Row data fields: (i) alignment rank as compound rank 'hitnum.alignmentnum', (ii) identifier, (iii) description, (iv) alignment 'bits', (v) alignment 'E', (vi) 'N'=1, (vii) 'hit' orientation.
TBLASTX
Input		Output		Orient		Assemble	Renumber	Filter
qry	hit	qry	hit	qry	hit
dna	dna	pro	pro	+	+	simple	yes	query orient
				+	-	simple	yes	query orient
				-	+	reverse	yes	query orient
				-	-	reverse	yes	query orient
For a given query orientation (+, -) derived from the query sequence numbering:
-hsp ranked		For each ranked hit, alignments corresponding to those in the ranking are selected for assembly having: (i) same expectation 'E', (ii) same score 'bits', (iii) same hit orientation. Row data fields: (i) rank as integral 'hitnum', (ii) identifier, (iii) description, (iv) ranked 'bits', (v) ranked 'E', (vi) 'N'=1, (vii) 'hit' orientation.
-hsp all		For each ranked hit, alignments are selected for assembly having: (i) same hit orientation. Row data fields: (i) rank as compound 'hitnum.alignmentnum', (ii) identifier, (iii) description, (iv) highest 'bits', (v) lowest 'E', (vi) total number of alignments 'N', (vii) 'hit' orientation.
-hsp discrete		For each ranked hit, every alignment is assigned its own row. Row data fields: (i) alignment rank as compound rank 'hitnum.alignmentnum', (ii) identifier, (iii) description, (iv) alignment 'bits', (v) alignment 'E', (vi) 'N'=1, (vii) 'hit' orientation.

[Home]

[FAQ]