TOC

• 1. Introduction
  1.1 What does blastall do?
  • 1.2 How does blastall find the alignment?
  • 1.3 Where can we find the blast program?
• 2. Installation and execution
  • 2.1 For PC platform running windows OS
    • 2.1.1 Downloading
    • 2.1.2 Installation
    • 2.1.3 Setup
    • 2.1.4 Execution
  • 2.2. For machines running Unix or Linux
    • 2.2.1 Downloading
    • 2.2.2 Installation
    • 2.2.3 Setup
    • 2.2.4 Execution
  • 2.3. For MacOSX
    • 2.3.1 Downloading and Installation
    • 2.3.2 Setup
    • 2.3.3 Execution
• 3. Program parameters for blastall
  • 3.1. Presentation format
  • 3.2. Individual command line options of blastall
• 4. General Usage
  • 4.1 Nucleotide vs nucleotide search using blastn
  • 4.2 Nucleotide vs protein search with blastx
  • 4.3 Protein vs protein search with blastp
  • 4.4 Protein vs nucleotide search with tblastn
  • 4.5 Nucleotide vs nucleotide search with tblastx
  • 4.6 Additional information and on searching with short queries
• 5. Additional information on scoring matrices and gap penalties
  • 5.1 Nucleotide scoring marices and their -G/-E values
  • 5.2 Protein scoring matrices and their -G/-E values
• 6. Feedback

BLAST is the acronym for "Basic Local Alignment Search Tool", which is a local alignment search tool first described in 1990 by Altschul et al [1]. NCBI started providing sequence alignment service to the public using BLAST in 1992, first through its blast email server (decommissioned in 2002) and later through the web (1997).

[1] Altschul SF, Gish W, Miller W, Myers EW and Lipman DJ (1990) J Mol Biol 215:403-410.

BLAST now refers to a family of different programs based on the same heuristics. These programs along with the accessory tools are made available to the public as a single binary package. Despite this expansion, blastall - the original program with enhenced features, remains to be the core of the BLAST package.

1.1 What does blastall do?

The all in "blastall" reveals some of the capability of this program. It can perform one of five different searches depending on the input specified through the -p option. The acceptable inputs and the type of search they performs are:

blastn   compares nucleotide queries to a nucleotide database
blastp   compares protein queries to a protein database
blastx   compares the translated products of nucleotide queries to a protein database
tblastn  compares protein queries to the translated products of a nucleotide database
tblastx  compares translational products of nucleotide queries to the translational products from a nucleotide database

1.2. How does blastall find the alignment?

BLAST finds the optimal alignment by using the "word matching" algorithm, in which BLAST does the search in several distinctive phases: 1) generating overlapping words from the input query, 2) scanning the database for word matches (hits), and 3) extending word hits to produce (local) alignments through three steps of extension.

During the first phase, BLAST breaks the input query into short overlapping segments, or "words," and stores them a hash table. BLAST takes those query words and scans the target database for initial matches in the second phase. The nucleotide BLAST algorithm looks for any single exact word match. The protein BLAST algorithm uses a scoring threshold cutoff to identify matches. In addition, protein BLAST algorithm also requires two word hits within a certain distance in order to proceed to the next step.

In the third phase, those initial matches or word hits are used as seeds to generate the alignments in through three extension step:

a. The first step is the un-gapped extension, in which BLAST extends the word hits in both directions to generate initial alignments without the introduction of gap.
b. The second step is the gapped extension, in which BLAST extends those initial un-gapped alignments further by introducing gaps in the alignments.
c. The last step is the final gapped extension with trace-back, in which BLAST does a final gapped extension attempt with trace-back to generate the actual alignments.

1.3 Where can we find the blast program?

Currently, the binaries for all these standalone blast programs and accessory tools are provided to the public as a single downloadable archive at ftp.ncbi.nlm.nih.gov/blast/executables/LATEST/. Most of the common platforms - combination of chipset and OS - are supported.

The installation of blast package can be divided into the following four steps:

1) Downloading
2) Extraction (installation)
3) Setup
4) Execution

2.1 For PC platform running Windows OS

2.1.1 Downloading

You can download the latest blast binary package for Windows from:

ftp.ncbi.nlm.nih.gov/blast/executables/LATEST/

blast-#.#.#-ia32-win32.exe

2.1.2 Installation

To keep all the relevant files/directories from the archive in one place, we need to fisrt create a new subdirectory under E:\ directory (or other directory of choice to you), named as blast-#.#.#, with #.#.# indicating the version number. For 2.2.13, we can name this subdirectory blast-2.2.13. Move the saved blast archive to this subdirectory and double click to extract the programs and other files. We will see a DOS terminal window open briefly with file names flashing by.

This installs the blast package and creates three subdirectories under the blast-2.2.13: bin, data, and doc. The bin subdiretory contains the following programs:

bl2seq.exe      blastall.exe     blastclust.exe      blastpgp.exe
copymat.exe     fastacmd.exe     formatdb.exe        formatrpsdb.exe
impala.exe      makemat.exe      megablast.exe       rpsblast.exe
seedtop.exe

2.1.3 Setup

To ensure the smooth execution of blast programs, we need to set up a BLAST configuration file, named ncbi.ini to instruct blast the location of the data directory and db directory. In the above setup, the path to data directory and db directory should be specified the follwing way:

[NCBI]
DATA=E:\blast-2.2.13\data  

[BLAST]
BLASTDB=E:\blast-2.2.13\db

• launch notepad
• copy paste the above section
• relace the path (E:\blast-2.2.13\) with the actual one for your setup
• name the file as ncbi.ini using "save as". Double quote the name to prevent Windows from adding the .txt extension
• move this resulted ncbi.ini file to your Windows folder

2.1.4 Execution

To run the program, you need first launch a DOS command prompt as depicted by the screenshot. DOS prompt is generally kept under:

Start ⇨ Program ⇨ Accessories ⇨ Command Prompt

Figure 2.1.4 Openning a DOS command prompt under window (XP)

This will open a command window. In this window, change the working directory to "E:\blast-2.2.13" by typing "E:" and enter key stroke, followed by "cd blast-2.2.13" and enter key stroke. Create the db directory using "mkdir db" command. Sample commad lines are listed in Table 2.1.4 for your reference.

C:\Documents and Settings\tao>e:
E:\>cd blast-2.2.13
E:\blast-2.2.13>mkdir db
E:\blast-2.2.13>dir
 Volume in drive E has no label.
 Volume Serial Number is EC96-7870
 
 Directory of E:\blast

12/14/2005  10:28 PM   <DIR>    .
12/14/2005  10:28 PM   <DIR>    ..
12/21/2005  12:05 AM   <DIR>    bin
12/04/2005  03:51 PM   <DIR>    data
12/04/2005  03:52 PM   <DIR>    db
12/04/2005  03:52 PM   <DIR>    doc
         6 Dir(s)  13,327,708,160 bytes free

E:\blast-2.2.13>

Change directory to E:\ drive
Change directory to blast-2.2.13 
Create db directory under blast-2.2.13
List files under blast-2.2.13 





System file
System file
bin directory
data directory
db directory
doc directory

To make windows aware of the location of blast programs, you need to add the path "E:\blast-2.2.13\bin\" to the PATH environment variable under "Start ⇨ Control Panel ⇨ System ⇨ Environment Variables ⇨ Path." Terminate the PATH string with a semicolon (;), then append "E:\blast2211\bin\" after it without the double quotes.

After this, we will be able to call the programs using their name followed by the appropriate option value pairs as input from any directory in the computer. For example, to call blastall, use its blastn subprogram, search against refseq_rna database with fasta_query.txt as query, and save the result in output.txt, we will use the following command line:

blastall -p blastn -d refseq_rna -i fasta_query.txt -o ouput.txt

2.2 For machines running Unix or Linux

2.2.1 Downloading

You can download the latest blast binary package from the same directory as for Windows:

ftp.ncbi.nlm.nih.gov/blast/executables/LATEST/

You need to select the appropriate archive for your platform (machine/OS combination) and save it to local disk by right click on the file and choose "Save target as ? within a browser. Or you can use FTP client to download it in binary form.

Note that NCBI does not support asymmetric protocol and requires that an FTP connection be made in PASSIVE mode. One should use ftp client with such capability, ncftp is one of such a client available.

2.2.2 Installation

To install, you need to inflate the downloaded archive first using the "gunzip -d archive.tar.gz" command line. The resulted tar file should be extracted using "tar -xvpf archive.tar" command line. Successful execution of the above two steps for the 2.2.13 archive will create a "blast-2.2.13" directory with the following subdirectories within it: bin, doc, data.

For better file management, you can also create an additional db subdirectory for database files using "mkdir db" command when you are under the blast-2.2.12 directory.

2.2.3 Setup

To set up, you should create a file named .ncbirc to store the configuration information. For example, if you keep the blast-2.2.13 directory under your home directory and you also created a db subdirectory within the blast-2.2.13, then you can specify the path to data and db directory the following way:

[NCBI]
DATA=~/blast-2.2.13/data

[BLAST]
BLASTDB=~/blast-2.2.13/db

2.2.4 Execution

If the path to blast-2.2.13/bin directory is not included in the PATH environment variable, you would need to call blast programs by appending the path before the program name. For example, under blast-2.2.13 directory, you can call the program using the following command line, where "./bin" instructs shell to look in the bin subdirectory under the current directory:

./bin/blastall -

2.3 For MacOSX

2.3.1 Downloading and Installation

Blast binary for MacOSX is under the same directory as for Windows and Unix platform. It is named with the following convention, with ppc32-macosx indicating the platform, e.g. powerPC based mac with OSX:

blast-#.#.#-ppc32-macosx.tar.gz  

Installation for Macintosh is essentially the same as under other Unix systems if one choose to download it using ftp client and performs the installation using command line. If you choose to download it using browser, StuffIt Expander will automatically expand the archive to install the folder, named blast-2.2.13 for current release, on the desktop with the preconfigured directory structure. After this, you should create a db folder and move the blast-2.2.13 folder to your own home directory, which generally resides at:

Computer icon ⇨ Home icon ⇨ icon with your name 

2.3.2 Setup

You need to launch Terminal program (mostly under the Utilities folder) to get a terminal prompt. By default, the working directory of terminal window is your home directory. Once there, you can create the .ncbirc file to put in the configuration information using pico editor and 'pico .ncbirc" command line (without quotes). The file should contain the following information:

[NCBI]
DATA=/Users/name/blast-2.2.13/data

[BLAST]
BLASTDB=/Users/name/blast-2.2.13/db

2.3.3 Execution

You can call the program using the instructions given for other Unix/Linux platforms given above. You can create a query file in this directory and search it against the database. For example if your working directory is blast-2.2.13, the following command will search sequences in an input query file "my_sequences" against a database called "nr". The output is to be saved in my_output:

./bin/blastall -p blastn -d nr -i my_sequence -o my_output

blastall is a command line program with no graphic user interface (GUI). We control the program through command line option switches issued in a terminal window. Those options instruct blastall what program function, query, and database to use. They also control the search sensitivity, format the result is return in, and to which file the result should be saved, etc. Typing "blastall -" followed by a return key stroke will display the blastall parameters along with short descriptions on each one.

3.1. Presentation format

The following section lists all the available options for blastall and provides functional explanation as well as usage examples. For clarity, each option is listed in its own table and all tables follow the same general format. Example provided for each option shows a valid option/value combination as they should appear in an actual command line. The options are based on the current 2.2.13 version. The first four options (-p, -i, -d, -o) are mandatory parameters for most searches.

3.2. Individual command line options of blastall

[2] Lupas et al. (1991). Science 252: 1162 - 1164
[3] Wilson et al. (1995). J. Gen. Virol. 76: 2923 - 2932. (Implemented by John Kuzio).

[4] Nuc Acid Res (2001) 29: 2994-3005.
[5] Bioinformatics (2005) 21: 902-911.

4.1 Nucleotide vs nucleotide search using blastn

Nucleotide vs nucleotide search can be used to identify the input sequences, find related sequences, map mRNA to its genomic counterpart, and map primers to their annealing target. The following command line searches the input nucleotide query in my_query file against the nt database using megablast algorithm with word size set to 56. The results are saved in my_output:

blastall -p blastn -i my_query -d nt -n T -W 56 -o my_output

4.2 Nucleotide vs protein search with blastx

This search involves the translation of the query and the actual alignment is performed at the protein level. It is useful in identify the potential protein product(s) that might be encoded by a nucleotide entry especially when the query nucleotide sequence still has certain errors. Also because the comparison is done at the more sensitive protein level, it is a useful search to identify homologous sequences that may escape the direct nucleotide search.

In the following command line, it searches nucleotide sequences in my_query against the protein database nr. The upper limit for returned description and alignments is set to 100 and the search results are saved in my_output:

blastall -p blastx -i my_query -d nr -v 100 -b 100 -o my_output

This search directly compares the input protein queries against a protein database. It is useful in finding other proteins that share sequence similarities to the input query. The matching sequences found by blastp can help identify the function of the input query. The following command line searches the input protein query in my_query file against the refseq_protein database. In attempt to identify the exact matches, if present in the databases, the filter function is inactivated. The result is saved in my_output:

blastall -p blastp -i my_query -d refseq_protein -F F -o my_output

Often the organism of interest only has a limited number of protein sequences available and the target gene of interest is not found there. One way to identify the gene sequence is through translated tblastn search against the nucletide sequences from the target organism. The query can be a known protein homolog from a well studied organism as human, mouse, or other model organisms.

The following search is an attempt to identity the gene sequences from Lactobacillus casei wgs that encodes the mismatch repair protein mutL. The query is Escherichia coli K12 mutL protein. The wgs search is restricted to the Lactobacillus casei entries using -l lactobacillus_gi [*].

blastall -i e_coli_mutL -d wgs -l lactobacillus_gi -p tblastn -o lacto_mutL.out

4.5 Nucleotide vs nucleotide search with tblastx

This search is generally reserved as a last resort when all the other searches (blastn, blastp, blastx, and tblastn) fail to return any useful information. Due to the computational intensity, it takes much more time to complete. In addition, tblastx searches tend to generate much more spurious hits and have a much higher noise to signal ratio. This makes the subsequent result interpretation very difficult and time consuming. It should be used with caution.

The following commandline attempts to find the mutL counterpart in Ferroplasma through tblastx search. The search is limited to Ferroplasma portion through the input Ferroplasma_gi file[**].

blastall -i e_coli_mutL.ORF -d wgs -l Ferroplasma_gi -p tblastx -o Ferroplasma_mutL.out

4.6 Additional information and on searching with short queries

The most commonly adjusted parameters are -F, -e, -b, -v, and -m. The remaining parameters are for more complex searches, which often require optimization that deviates from the default settings. For example, to try to get longer alignment, your can try increasing the X-dropoff values specified by -X, -y, and -Z parameters.

As mentioned in the beginning, -e, -b, -v cutoff limits are applied at each of the three alignment steps. The side-effect of stringent settings in those parameters is that certain HSPs may not be included in the final result since the HSPs from ungapped extension step fall below the cutoff and was not carried to the gapped extension steps.

When BLAST searching with short query sequences, there will not be reliable way to gauge the statistical significance of the matches due the short nature and biased composition of the query. To make this type of searches working, special option/value pairs will be needed. For short nulcoetide query, we recommend adding the following option/value pairs to the command line:

-F F -e 1000 -W 7

-F F -e 20000 -W 2 -M PAM30

5.1 Nucleotide scoring marices and their -G/-E values

Nucleodite blast searches through megablast, blastall, and bl2seq have until now allowed users to select arbitrary gap existence and extension penalties for a blastn type search. This has been convenient for users but has led to the unfortunate situation that searches with some parameter sets were significantly overestimating the statistical significance of matches. The parameters that might cause an issue here are -r, -q, -G, and -E.

To address this, the proper statistical parameters for a number of -r /-q/-G/-E values have been calculated starting from version 2.2.13. Note that above a certain gap existence and extension penalty any value is permitted, as the statistics for ungapped searches can be used. These are marked as "ungapped threshold" below.

5.2 Protein scoring marices and their -G/-E values

NCBI BLAST programs support five protein scoring matrices: BLOSUM45, BLOSUM62, BLOSUM80, PAM30, and PAM70. To ensure correct statisitcal evaluation of the found matches, only a limitted number of -G/-E sets. These matrices with their supported -G/-E values are summarized in the table below.

For questions and comments on this document and BLAST in general, please send them to:

blast-help@ncbi.nlm.nih.gov

info@ncbi.nlm.nih.gov