Saturated BLAST User's Manual

ISS is a strategy for recognizing distant homologues using transitive sequences. This idea is when the similarity between two remotely homologous sequences can not be detected by normal sequence comparison, if there is an intermediate sequence with significant alignment scores to both of them, their similarity can still be established.

ISS and its extension, multiple ISS, which applies more than one intermediate steps, have been proved to be sensitive and practical. However, the brute-force search, which repetitively runs database searching, is time-consuming and hard to be done in an automated way. Saturated BLAST is a software with graphic user interface, it can perform the iterated multiple intermediate sequence search more efficiently and automatically.

The Saturated BLAST package was developed under LINUX system using Perl as the programming language. Starting with a single query or a set of related sequences, Saturated BLAST runs a BLAST search, organizes the output, identifies representative as search seeds, and then repetitively takes these new seeds as queries for next generation of BLAST searches. The friendly graphic user interface and the built-in BLAST result parser, multiple alignment tools and clustering algorithms provide an easy way to edit, visualize, analyze, monitor and control the search.

  Weizhong Li, Frederic Pio, Krzysztof Pawlowski & Adam Godzik.
  Saturated BLAST: An automated multiple intermediate sequence search
  used to detect distant homology. Bioinformatics (2000) in press

  use Tk;
  use LWP::UserAgent;
  use LWP::Simple;
  use HTTP::Request::Common;

  gunzip < xblast_version_number.tar.gz | tar xvf -

  my $XBLAST_ROOT = "/usr/local/bin/xblast";

  my $blast_root = "/data/ncbi"; 
  $ENV{"NCBI"} = $blast_root; 
  $ENV{"BLASTDB"} = "$blast_root/db"; 
  $ENV{"BLASTMAT"} = "$blast_root/data"; 

• Seed: sequences found in database search and used as queries in later searches.
• Parent: A seed is the parent of all the sequences found with it.
• Level: The sequences input by user has level of 0, its children have level of 1, and its children's children have level of 2. In other words, the level of a protein is the intermediate step from top to itself.
• Break point: A seed can be marked as a break point, where the Saturated BLAST search will stop and wait for user's input.
• Hit no: The the number of parents for a sequence. Different seeds can often find same sequence, so a sequence may have more than one parents, althrough Saturated BLAST only display the first one.

• Redundancy filter. It is used to prevent the addition of redundant sequences. Different queries can often find same sequence fragment. For every new sequence, if there is another sequence with same ID, and if the sequence overlap between them is bigger than a threshold, this sequence is removed.
• Low significance filter. It is used to remove the sequences with bad expect values, low sequence identities or short lengthes.
• Keywords filter. It is used to remove or keep sequences according to some keyword patterns. The pattern is compared to the annotation from BLAST search.
• Smart filter. The program can remember the sequcnes deleted by user, and put the IDs of them in a list. When this function is enabled, the sequences whose ID can be found in the list are deleted.

• Multiple alignment of single BLAST search. Since the alignments between all the hit sequence with query are printed in the single BLAST output, the multiple alignment of them can be directly derived.
• Multiple alignment of MISS search chain. The alignment of individual sequence, the orginal query, and all the intermediate sequence between them can be built by recalling the intermediate steps.
• Alignment of any two sequences. If two sequences have same ancestor, the alignment between them can be calculated by aligning them to the ancestor. Otherwise, they can be aligned by the built-in dynamic program tool.
• Multiple alignment of any group of sequences. This alignment can be drived if they have same ancestor as before. But if the protein family is very huge, the alignment could be terriable.

  S = s - [ln(Kmn)/ lambda ]

• no, a serial number assigned to each sequence.
• gi, the genebank gi identifier.
• description, the annotation of sequence.
• database, the database (undisplayed as default).
• temporary, a temporary field used by Saturated BLAST (undisplayed as default).
• parent, the serial number of its parent.
• seed, a red 'x' is marked for active seed, a normal 'x' is marked for used seed.
• break, a red 'x' is marked for a break point, where the Saturated BLAST will stop.
• level, the level of this sequence.
• hit no, the hit no of this sequence.
• score, the alignment of this sequence.
• expect, the expect value of thie sequence.
• identity, the sequence identity of this sequence.
• query range, the range of query sequence in the BLAST alignment.
• hit range, the range of itself in the BLAST alignment.
• query sequence, the segment of query sequence in the BLAST alignment.
• hit sequence, the segment of this sequence in the BLAST alignment.
• Aligned sequence, the sequence aligned its full-length parent without gaps in parent.

• double click on serial no or description will open a message window of this sequence.
• double click on gi number will refresh a netscape window to connect NCBI server and display the annotation of this sequence. This require a existing netscape window on your desktop. If nothing happen after clicking, check whether you have open a netscape window or whether the action is performed on other desktop.
• double click on parent will open a message window of this sequence's parent.
• click on the non-text part of a sequence will select or unselect it. the selected sequences have light yellow background.
• click on the non-text part of a sequence while pressing shift key will select a range of sequences.

• File open

• File new

• File save

• File save as

• File export

• File quit

• Edit delete selected

• Edit delete with children

• Edit delete with parent

• Edit undelete

• Edit clear deletion

• Edit insert sequences

• Edit search

• Max iteration number is the max allowed level. The program will stop and a message will printed in the status bar if this value is reached.
• Max BLAST search number is the max allowed number of BLAST searches. The program will stop and a message will printed in the status bar if this value is reached.
• Batch work size is the max allowed number of BLAST searches in one background batch.

• Max expect of hits: sequences with expect values higher than this value will be deleted from BLAST results.
• Min length of hits: sequences with length shorter than this value will be deleted.
• Min sequence identity of hits: sequence will be deleted if the sequence identity with their parent is lower then this value.
• Threshold overlap percent: for a new sequence B from BLAST results, if its id (gi number) is same with another sequence A of Saturated BLAST, a overlap percent is calculated as the overlaped length between a and b divided by the length of B. If this value is larger than the "Threshold overlap percent", sequence B is considered as same as A, and be deleted.
• Keywords to include: only sequences match these keywords from BLAST results will added by the program.
• Keywords include pattern: the logical pattern for "Keywords to include".
• Keywords to exclude: sequences matching these keywords will be deleted from BLAST results.
• Keywords exclude pattern: the logical pattern for "Keywords to exclude".
• Filter mode: it is for the "Threshold overlap percent", the default value is strict, and the default behavior is described above. But when it is set to greddy, if overlap between B and A is greater than threshold, but A is not so significant to be seed, B will not be deleted.

• Max expect for seed: sequences with expect values higher than this value will not be selected as seed.
• Min length of seed: sequences with length shorter than this value will not be selected as seeds.
• Min sequence identity of seed: sequence will not be selected as seeds if the sequence identity with their parent is lower then this value.
• Cluster threshold: the threshold of sequence identity to cluster the sequence from BLAST output, and the longest sequences from each cluster are selected as seeds.

• Stop when match words: the program will stop if these words are found in annotations of sequences.
• Logical pattern: The logical pattern for above function.

• Set seed

• Set unset seed

• Set clear seed

• Set default seed

• Set reuse seed

• Set break point on

• Set break point off

• Set clear break point

• Set smart filter

• Set show filter list

• Set export/import filter list

• Set clear filter list

• Select all

• Select clear

• Select inverse

• Select logic

• Display refresh

• Display all

• Display seed

• Display selected

• Display sort by

• Display sort mode

• Display show (gi, hit no, and so on)

• Tool BLAST alignment

• Tool linked alignment

• Menu File -> refresh refreshes the alignment.
• Menu File -> save saves the alignment to a text file in FASTA format.
• Menu File -> close closes this window.
• Menu Configuration -> show gap in master switches on or off the display of gaps in master sequences.
• Menu Configuration -> show what? allowes user to define what sequences to be displayed. The selection tool and display tool from main window can be used when the window is open.

• Tool view log

• Tool align 2 sequences

• Tool cluster selected

• Menu File -> refresh refreshes results.
• Menu File -> save saves the cluster results into a text file.
• Menu File -> close closes this window.
• Menu Set -> select representatives will select a representative from each cluster, this selection will replace the current setting in the main window.
• Menu Set -> mark clstr no assignes the cluster number to the temporary filed of clustered sequences. User then can use display tool to view this number in main display window or to sort results.
• Menu clear pairwise score delete all saved pairwise alignment scores or identities.
• Menu use sequence identity use the sequence identity rather than alignment score for clustering.

• Help help

• Help manual (offical, local)

• Button next run BLAST search using next seed.
• Button cont contiunely run BLAST searches until break point is met or no new seed is available.
• Button jump to run BLAST searching using next selected seed.
• Button send batch send a batch of BLAST searches in background. The size of batch is defined by menu Set->parameter. These jobs are then hold in a waiting list. While the waiting list is full (has the same size of defined batch), no more jobs of background can be submitted.
  NOTE! if you sumbit jobs to NCBI server, please don't define big batch size, because NCBI is a public server.
• Button get batch will check waiting list whether the background jobs are finished. If some jobs are finished, they are parsed and added into Saturated BLAST. If some jobs are still running, program will return and leave them in the waiting list.
• Button run batch will run batch BLAST automaticly, it will stop if break point is met or no new seed is available.
• Button stop will send a stop sign to program, and the BLAST search will stop after receives this signal.