The rules used in MView for selecting and tiling HSPs

A BLAST run comprises, among other sections, a 'ranking' of hits with one-line summaries including scoring information, followed by the search database 'hits' themselves. Each hit may contain one or more HSPs (ungapped BLAST) or gapped alignments (gapped BLAST), not all of which may be reported in the ranking.

Some control over which HSPs are to be processed by MView is provided by the command line option -hsp mode. There are three choices:

ranked

This is the default behaviour. Only those HSPs or alignments that contributed to the ranked hit are selected. These are tiled into a single alignment row per hit.

all

All HSPs or alignments are selected and tiled into a single alignment row per hit.

discrete

All HSPs or alignments are selected but each is assigned its own row in the resulting stacked alignment.

A more detailed and specific description of the selection rules is given for each BLAST program for each series recognised by MView:

• ungapped NCBI BLAST 1.4
• gapped WU-BLAST 2.0
• ungapped NCBI BLAST 2.0 and PSI-BLAST

General notes

Assemble (simple)

Given a list of fragments each numbered from low to high along the query strand (ie., conventional orientation): stack these at these positions along the query template.

Assemble (reverse)

Given a list of fragments each numbered from high to low along the query strand (ie., individually reversed): reverse each fragment, stack onto the template, reverse the template, which recovers the original numbering high to low and preserves intra- and inter-fragment order.

alignmentnum

The ordinal number (starting at 1) of a single HSP within a given hit (starting afresh with each hit) as encountered in the raw BLAST output.

bits

NCBI BLAST 2.0 hit ranking rounds HSP scores which are floating point to the nearset integer, so MView takes this into account when selecting HSPs.

Filter

Separate MView output must be produced for varous BLAST hit properties, such as PSI-BLAST search cycle, BLASTN query strand orientation, etc.

hitnum

The ordinal number (starting at 1) of a search hit as encountered in the ranking section of the raw BLAST output.

N

A count of HSPs which is context dependent in MView. In raw BLAST output it is the number of HSPs deemed by BLAST to belong in a coherent set of alignment fragments, but the MView HSP processing options (ranked, all, discrete) change the precise meaning.

P(N)

The BLAST alignment p-value associated with a set of N HSPs.

p-value

BLAST 1.4 series reports relatively large p-vales in floating point format rather than scientific notation. The original HSP p-value may be rounded to 2 decimal places in the hit ranking, so comparison of these values by MView when selecting HSPs takes this into account, which could lead to over-selection of HSPs on occasion.

Renumber

Given a (start,stop) pair in nucleotide units, the corresponding nearest values in amino acid units are given by:

  (start', stop') = ( int((start+2)/3), int(stop/3) )  

This renumbering is applied to each fragment's positions along its parent query strand to determine the query template positions in amino acid units.

score

The BLAST alignment score.