BLAT Suite Program Specifications and User Guide


Blat produces two major classes of alignments: at the DNA level between two sequences that are of 95% or greater identity, but which may include large inserts, and at the protein or translated DNA level between sequences that are of 80% or greater identity and may also include large inserts.  The output of BLAT is flexible.  By default it is a simple tab-delimited file which describes the alignment, but which does not include the sequence of the alignment itself.  Optionally it can produce BLAST and WU-BLAST compatable output as well as a number of other formats.


The main programs in the blat suite are:

Building an index of the genome typically takes 10 or 15 minutes.  Typically for interactive applications one uses gfServer to build a whole genome index.  At that point gfClient or webBlat can align a single query within few seconds.  If one is aligning a lot of sequences in a batch mode then blat can be more efficient, particularly if run on a cluster of computers.  Each blat run is typically done against a single chromosome, but with a large number of query sequences.


Other programs in the blat suite are:


In addition you may be interested in the following programs which are not part of the BLAT suite:


Running  the  Programs:

The command line options of each of the programs is described below. Similar summaries of usage are printed when a command is run with no arguments.  See the next section for info on installing webBlat.




blat - Standalone BLAT sequence search command line tool


   blat database query [-ooc=11.ooc] output.psl


   database and query are each either a .fa , .nib or .2bit file,

   or a list these files one file name per line.

   -ooc=11.ooc tells the program to load over-occurring 11-mers from

               and external file.  This will increase the speed

               by a factor of 40 in many cases, but is not required

   output.psl is where to put the output.

   Subranges of nib and .2bit files may specified using the syntax:






   With the second form, a sequence id of file:start-end will be used.


   -t=type     Database type.  Type is one of:

                 dna - DNA sequence

                 prot - protein sequence

                 dnax - DNA sequence translated in six frames to protein

               The default is dna

   -q=type     Query type.  Type is one of:

                 dna - DNA sequence

                 rna - RNA sequence

                 prot - protein sequence

                 dnax - DNA sequence translated in six frames to protein

                 rnax - DNA sequence translated in three frames to protein

               The default is dna

   -prot       Synonymous with -t=prot -q=prot

   -ooc=N.ooc  Use overused tile file N.ooc.  N should correspond to

               the tileSize

   -tileSize=N sets the size of match that triggers an alignment. 

               Usually between 8 and 12

               Default is 11 for DNA and 5 for protein.

   -stepSize=N spacing between tiles. Default is tileSize.

   -oneOff=N   If set to 1 this allows one mismatch in tile and still

               triggers an alignments.  Default is 0.

   -minMatch=N sets the number of tile matches.  Usually set from 2 to 4

               Default is 2 for nucleotide, 1 for protein.

   -minScore=N sets minimum score.  This is the matches minus the

               mismatches minus some sort of gap penalty.  Default is 30

   -minIdentity=N Sets minimum sequence identity (in percent).  Default is

               90 for nucleotide searches, 25 for protein or translated

               protein searches.

   -maxGap=N   sets the size of maximum gap between tiles in a clump.  Usually

               set from 0 to 3.  Default is 2. Only relevent for minMatch > 1.

   -noHead     suppress .psl header (so it's just a tab-separated file)

   -makeOoc=N.ooc Make overused tile file. Target needs to be complete genome.

   -repMatch=N sets the number of repetitions of a tile allowed before

               it is marked as overused.  Typically this is 256 for tileSize

               12, 1024 for tile size 11, 4096 for tile size 10.

               Default is 1024.  Typically only comes into play with makeOoc.

               Also affected by stepSize. When stepSize is halved repMatch is

               doubled to compensate.

   -mask=type  Mask out repeats.  Alignments won't be started in masked region

               but may extend through it in nucleotide searches.  Masked areas

               are ignored entirely in protein or translated searches. Types are

                 lower - mask out lower cased sequence

                 upper - mask out upper cased sequence

                 out   - mask according to database.out RepeatMasker .out file

                 file.out - mask database according to RepeatMasker file.out

   -qMask=type Mask out repeats in query sequence.  Similar to -mask above but

               for query rather than target sequence.

   -repeats=type Type is same as mask types above.  Repeat bases will not be

               masked in any way, but matches in repeat areas will be reported

               separately from matches in other areas in the psl output.

   -minRepDivergence=NN - minimum percent divergence of repeats to allow

               them to be unmasked.  Default is 15.  Only relevant for

               masking using RepeatMasker .out files.

   -dots=N     Output dot every N sequences to show program's progress

   -trimT      Trim leading poly-T

   -noTrimA    Don't trim trailing poly-A

   -trimHardA  Remove poly-A tail from qSize as well as alignments in

               psl output

   -fastMap    Run for fast DNA/DNA remapping - not allowing introns,

               requiring high %ID

   -out=type   Controls output file format.  Type is one of:

                   psl - Default.  Tab separated format, no sequence

                   pslx - Tab separated format with sequence

                   axt - blastz-associated axt format

                   maf - multiz-associated maf format

                   sim4 - similar to sim4 format

                   wublast - similar to wublast format

                   blast - similar to NCBI blast format

                   blast8- NCBI blast tabular format

                   blast9 - NCBI blast tabular format with comments

   -fine       For high quality mRNAs look harder for small initial and

               terminal exons.  Not recommended for ESTs

   -maxIntron=N  Sets maximum intron size. Default is 750000

   -extendThroughN - Allows extension of alignment through large blocks of N's



Here are some blat settings for common usage scenarios:


1) Mapping ESTs to the genome within the same species


2) Mapping full length mRNAs to the genome in the same species

    -ooc=11.ooc -fine -q=rna

3) Mapping ESTs to the genome across species

    -q=dnax -t=dnax

4) Mapping mRNA to the genome across species

    -q=rnax -t=dnax

5) Mapping proteins to the genome

    -q=prot -t=dnax

6) Mapping DNA to DNA in the same species

    -ooc=11.ooc -fastMap

7) Mapping DNA from one species to another species

    -q=dnax -t=dnax

    When mapping DNA from one species to another the

    query side of the alignment should be cut up into chunks

    of 25kb or less for best performance.



gfServer - Make a server to quickly find where DNA occurs in genome.

To set up a server:

   gfServer start host port file(s)

   Where the files are in .nib or .2bit format

To remove a server:

   gfServer stop host port

To query a server with DNA sequence:

   gfServer query host port probe.fa

To query a server with protein sequence:

   gfServer protQuery host port probe.fa

To query a server with translated dna sequence:

   gfServer transQuery host port probe.fa

To process one probe fa file against a .nib format genome (not starting server):

   gfServer direct probe.fa file(s).nib

To figure out usage level

   gfServer status host port

To get input file list

   gfServer files host port


   -tileSize=N size of n-mers to index.  Default is 11 for nucleotides, 4 for

               proteins (or translated nucleotides).

   -stepSize=N spacing between tiles. Default is tileSize.

   -minMatch=N Number of n-mer matches that trigger detailed alignment

               Default is 2 for nucleotides, 3 for protiens.

   -maxGap=N   Number of insertions or deletions allowed between n-mers.

               Default is 2 for nucleotides, 0 for protiens.

   -trans  Translate database to protein in 6 frames.  Note: it is best

           to run this on RepeatMasked data in this case.

   -log=logFile keep a log file that records server requests.

   -seqLog    Include sequences in log file (not logged with -syslog)

   -syslog    Log to syslog

   -logFacility=facility log to the specified syslog facility - default local0.

   -mask      Use masking from nib file.

   -repMatch=N Number of occurrences of a tile (nmer) that trigger

               repeat masking the tile. Default is 1024.

   -maxDnaHits=N Maximum number of hits for a dna query that are sent from the

               server. Default is 100.

   -maxTransHits=N Maximum number of hits for a translated query that are

               sent from the server. Default is 200.

   -maxNtSize=N Maximum size of untranslated DNA query sequence

               Default is 40000

   -maxAsSize=N Maximum size of protein or translated DNA queries

               Default is 8000

   -canStop If set then a quit message will actually take down the




gfClient - A client for the genomic finding program


   gfClient host port nibDir in.fa out.psl


   host is the name of the machine running the gfServer

   port is the same as you started the gfServer with

   nibDir is the path of the nib files relative to the current dir

       (note these are needed by the client as well as the server)

   in.fa a fasta format file.  May contain multiple records

   out.psl where to put the output


   -t=type     Database type.  Type is one of:

                 dna - DNA sequence

                 prot - protein sequence

                 dnax - DNA sequence translated in six frames to protein

               The default is dna

   -q=type     Query type.  Type is one of:

                 dna - DNA sequence

                 rna - RNA sequence

                 prot - protein sequence

                 dnax - DNA sequence translated in six frames to protein

                 rnax - DNA sequence translated in three frames to protein

   -dots=N   Output a dot every N query sequences

   -nohead   Suppresses psl five line header

   -minScore=N sets minimum score.  This is twice the matches minus the

               mismatches minus some sort of gap penalty.  Default is 30

   -minIdentity=N Sets minimum sequence identity (in percent).  Default is

               90 for nucleotide searches, 25 for protein or translated

               protein searches.


   -out=type   Controls output file format.  Type is one of:

                   psl - Default.  Tab separated format without actual sequence

                   pslx - Tab separated format with sequence

                   axt - blastz-associated axt format

                   maf - multiz-associated maf format

                   wublast - similar to wublast format

                   blast - similar to NCBI blast format

   -maxIntron=N  Sets maximum intron size. Default is 750000




webBlat generally is not run from the command line.  See ‘Setting Up webBlat instructions below` for information on this program.



faToTwoBit - Convert DNA from fasta to 2bit format


   faToTwoBit in.fa [in2.fa in3.fa ...] out.2bit


   -noMask - Ignore lower-case masking in fa file.



twoBitToFa - Convert all or part of .2bit file to fasta


   twoBitToFa input.2bit output.fa


   -seq=name - restrict this to just one sequence

   -start=X  - start at given position in sequence (zero-based)

   -end=X - end at given position in sequence (non-inclusive)



faToNib - Convert from .fa to .nib format


   faToNib in.fa out.nib



nibFrag - Extract part of a nib file as .fa


   nibFrag file.nib start end strand out.fa



pslPretty - Convert PSL to human readable output


   pslPretty in.psl target.lst query.lst pretty.out


   -axt - save in Scott Schwartz's axt format

   -dot=N Put out a dot every N records

   -long - Don't abbreviate long inserts


It's a really good idea if the psl file is sorted by target if it contains multiple targets.  Otherwise this will be very very slow.   The target and query lists can either be fasta files, nib files, or a list of fasta and/or nib files one per line.  Currently this only handles nucleotide based psl files.



pslSort - merge and sort psCluster .psl output files


  pslSort dirs[1|2] outFile tempDir inDir(s)

This will sort all of the .psl files in the directories

inDirs in two stages - first into temporary files in tempDir

and second into outFile.  The device on tempDir needs to have

enough space (typically 15-20 gigabytes if processing whole genome)

  pslSort g2g[1|2] outFile tempDir inDir(s)

This will sort a genome to genome alignment, reflecting the

alignments across the diagonal.


Adding 1 or 2 after the dirs or g2g will limit the program to

only the first or second pass repectively of the sort



   -verbose=N Set verbosity level, higher for more output. Default 1


Note for huge files pslSort will run out of memory.  The unix sort command

  sort –k 10 *.psl > sorted.psl

may be preferable in these situations, though the psl header lines should

be removed or avoided with the –noHead option to blat.



pslReps - analyse repeats and generate genome wide best

alignments from a sorted set of local alignments


    pslReps in.psl out.psl out.psr

where in.psl is an alignment file generated by psLayout and

sorted by pslSort, out.psl is the best alignment output

and out.psr contains repeat info


    -nohead don't add PSL header

    -ignoreSize Will not weigh in favor of larger alignments so much

    -noIntrons Will not penalize for not having introns when calculating

              size factor

    -singleHit  Takes single best hit, not splitting into parts

    -minCover=0.N minimum coverage to output.  Default is 0.

    -ignoreNs Ignore 'N's when calculating minCover.

    -minAli=0.N minimum alignment ratio

               default is 0.93

    -nearTop=0.N how much can deviate from top and be taken

               default is 0.01

    -minNearTopSize=N  Minimum size of alignment that is near top

               for aligmnent to be kept.  Default 30.

    -coverQSizes=file Tab-separate file with effective query sizes.

                     When used with -minCover, this allows polyAs

                     to be excluded from the coverage calculation


Setting Up webBlat



Installing A Web-Based Blat Server involves four major steps:

 1) Creating sequence databases.

 2) Running the gfServer program to create in-memory indexes of the databases.

 3) Editing the webBlat.cfg file to tell it what machine and port the gfServer(s)

    are running on, and optionally customizing the webBlat appearance to users.

 4) Copying the webBlat executable and webBlat.cfg to a directory where the web server can

    execute webBlat as a CGI.




You create databases with the program faToTwoBit. Typically you'll create

a separate database for each genome you are indexing.  Each database can

contain up to four billion bases of sequence in an unlimited number of

records.  The databases for webPcr and webBlat are identical.


The input to faToTwoBit is one or more fasta format files each of which

can contain multiple records.  If the sequence contains repeat sequences,

as is the case with vertebrates and many plants, the repeat sequences can

be represented in lower case and the other sequence in upper case.  The gfServer

program can be configured to ignore the repeat sequences.  The output of

faToTwoBit is a file which is designed for fast random access and efficient

storage.  The output files store four bases per byte.  They use a small amount

of additional space to store the case of the DNA and to keep track of runs of

N's in the input.  Non-N ambiguity codes such as Y and U in the input sequence

will be converted to N.


Here's how a typical installation might create a mouse and a human genome database:

    cd /data/genomes

    mkdir twoBit

    faToTwoBit human/hg16/*.fa twoBit/hg16.2bit

    faToTwoBit mouse/mm4/*.fa twoBit/mm4.2bit

There's no need to put all of the databases in the same directory, but it can

simplify bookkeeping. 


The databases can also be in the .nib format which was used with blat and

gfClient/gfServer until recently.  The .nib format only packed 2 bases per

byte, and could only handle one record per nib file.  Recent versions of blat

and related programs can use .2bit files as well.




The gfServer program creates an in-memory index of a nucleotide sequence database. 

The index can either be for translated or untranslated searches.  Translated indexes

enable protein-based blat queries and use approximately two bytes per unmasked base

in the database.  Untranslated indexes are used nucleotide-based blat queries as well

as for In-silico PCR.  An index for normal blat uses approximately 1/4 byte per base.  For

blat on smaller (primer-sized) queries or for In-silico PCR a more thorough index that

requires 1/2 byte per base is recommended.  The gfServer is memory intensive but typically

doesn not require a lot of CPU power.  Memory permitting multiple gfServers can be

run on the same machine. 


A typical installation might go:

    ssh bigRamMachine

    cd /data/genomes/twoBit

    gfServer start bigRamMachine 17779 hg16.2bit &

    gfServer -trans -mask start bigRamMachine 17778 hg16.2bit &

the -trans flag makes a translated index.   It will take approximately

15 minutes to build an untranslated index, and 45 minutes to build a

translate index.  To build an untranslated index to be shared with

In-silico PCR do

    gfServer -stepSize=5 bigRamMachine 17779 hg16.2bit &

This index will be slightly more sensitive, noticeably so for small query sequences,

with blat.




The webBlat.cfg file tells the webBlat program where to look for gfServers and

for sequence.  The basic format of the .cfg file is line oriented with the

first word of the line being a command.  Blank lines and lines starting with #

are ignored.  The webBlat.cfg and webPcr.cfg files are similar. The webBlat.cfg

commands are:

   gfServer  -  defines host and port a (untranslated) gfServer is running on, the

                                associated sequence directory, and the name of the database to display in

                the webPcr web page.

   gfServerTrans - defines location of a translated server.

   background - defines the background image if any to display on web page

   company    - defines company name to display on web page

   tempDir    - where to put temporary files.  This path is relative to where the web

                server executes CGI scripts.  It is good to remove files that haven't

                                been accessed for 24 hours from this directory periodically,

                                via a cron job or similar mechanism.

The background and company commands are optional.  The webBlat.cfg file must

have at least one valid gfServer or gfServerTrans line, and a tempDir line.

.  Here is a webBlat.cfg file that you

might find at a typical installation:


company Awesome Research Amalgamated

background /images/dnaPaper.jpg

gfServer bigRamMachine 17778 /data/genomes/2bit/hg16.2bit Human Genome

gfServer bigRamMachine 17779 /data/genomes/2bit/hg16.2bit Human Genome

gfServer mouseServer 17780 /data/genomes/2bit/mm4.2bit Mouse Genome

gfServer mouseServer 17781 /data/genomes/2bit/mm4.2bit Mouse Genome

tempDir ../trash




The details of this step vary highly from web server to web server.  On

a typical Apache installation it might be:

     ssh webServer

     cd kent/webBlat

     cp webBlat webBlat.cfg /usr/local/apache/cgi-bin

     mkdir /usr/local/apache/trash

     chmod 777 /usr/local/apache/trash

assuming that you've put the executable and config file in kent/webBlat.

The program will create some files in the trash directory.  It is good to

periodically clean out old files from this directory.   On Mac OS-X instead

you might do:

     cp webBlat webBlat.cfg /Library/WebServer/CGI-Executables

     mkdir /Library/WebServer/trash

     chmod 777 /Library/WebServer/trash

Unless you are administering your own computer you will likely need to ask your

local system administrators for help with this part of the webBlat installation.


File Formats


.psl files

A .psl file describes a series of alignments in a dense easily parsed text format.  It begins with a five line header which describes each field.  Following this is one line for each alignment with a tab between each field.  The fields are describe below in  a format suitable for many relational databases.

    matches int unsigned ,           # Number of bases that match that aren't repeats

    misMatches int unsigned ,        # Number of bases that don't match

    repMatches int unsigned ,        # Number of bases that match but are part of repeats

    nCount int unsigned ,            # Number of 'N' bases

    qNumInsert int unsigned ,        # Number of inserts in query

    qBaseInsert int unsigned ,       # Number of bases inserted in query

    tNumInsert int unsigned ,        # Number of inserts in target

    tBaseInsert int unsigned ,       # Number of bases inserted in target

    strand char(2) ,                 # + or - for query strand, optionally followed by + or – for target strand

    qName varchar(255) ,             # Query sequence name

    qSize int unsigned ,             # Query sequence size

    qStart int unsigned ,            # Alignment start position in query

    qEnd int unsigned ,              # Alignment end position in query

    tName varchar(255) ,             # Target sequence name

    tSize int unsigned ,             # Target sequence size

    tStart int unsigned ,            # Alignment start position in target

    tEnd int unsigned ,              # Alignment end position in target

    blockCount int unsigned ,        # Number of blocks in alignment. A block contains no gaps.

    blockSizes longblob ,            # Size of each block in a comma separated list

    qStarts longblob ,               # Start of each block in query in a comma separated list

    tStarts longblob ,               # Start of each block in target in a comma separated list

In general the coordinates in psl files are “zero based half open.” The first base in a sequence is numbered zero rather than one. When representing a range the end coordinate is not included in the range. Thus the first 100 bases of a sequence are represented as 0-100, and the second 100 bases are represented as 100-200. There is a another little unusual feature in the .psl format. It has to do with how coordinates are handled on the negative strand. In the qStart/qEnd fields the coordinates are where it matches from the point of view of the forward strand (even when the match is on the reverse strand). However on the qStarts[] list, the coordinates are reversed.


Here's an example of a 30-mer that has 2 blocks that align on the minus strand and 2 blocks on the plus strand (this sort of stuff happens in real life in response to assembly errors sometimes).

0         1         2         3 tens position in query
0123456789012345678901234567890 ones position in query
            ++++          +++++ plus strand alignment on query
    --------    ----------      minus strand alignment on query

Plus strand:
     qStart 12 qEnd 31 blockSizes 4,5 qStarts 12,26
Minus strand:
     qStart 4 qEnd 26 blockSizes 10,8 qStarts 5,19

Essentially the minus strand blockSizes and qStarts are what you would get if you reverse complemented the query.However the qStart and qEnd are non-reversed. To get from one to the other:
     qStart = qSize - revQEnd
     qEnd = qSize - revQStart

.2bit files

A .2bit file can store multiple DNA sequence (up to 4 gig total) in a compact randomly accessible format.  The two  bit files contain masking information as well as the DNA itself.   The file begins with a 16 byte header containing the following fields:

1)       signature – the number 0x1A412743 in the architecture of the machine that created the file.

2)       version – zero for now. Readers should abort if they see a version number higher than 0.

3)       sequenceCount – the number of sequences in the file

4)       reserved – always zero for now.

All fields are 32 bits unless noted. If the signature value is not as given, the reader program should byte swap the signature and see if the swapped version matches.  If so all multiple-byte entities in the file will need to be byte-swapped.  This enables these binary files to be used unchanged on different architectures.


The header is followed by a file index.  There is one entry in the index for each sequence.   Each index entry contains three fields:

1)       nameSize – a byte containing the length of the name field

2)       name – this contains the sequence name itself, and is variable length depending on nameSize.

3)       offset – 32 bit offset of the sequence data relative to the start of the file


The index is followed by the sequence records.  These contain 9 fields:

1)       dnaSize – number of bases of DNA in the sequence.

2)       nBlockCount – the number of blocks of N’s in the file (representing unknown sequence).

3)       nBlockStarts – a starting position for each block of N’s

4)       nBlockSizes – the size of each block of N’s

5)       maskBlockCount – the number of masked (lower case) blocks

6)       maskBlockStarts – starting position for each masked block

7)       maskBlockSizes – the size of each masked block

8)       packedDna – the dna packed to two bits per base as so: 00 – T,  01 – C,  10 – A, 11 – G.  The first base is in the most significant 2 bits byte, and the last base in the least significant 2 bits, so that the sequence TCAG would be represented as 00011011.  The packedDna field will be padded with 0 bits as necessary  so that it takes an even multiple of 32 bit in the file, as this improves i/o performance on some machines.



.nib files

A .nib file describes a DNA sequence packing two bases into each byte. Each nib file contains only a single sequence.  A nib file begins with a 32 bit signature which is 0x6BE93D3A in the architecture of the machine that created the file, and possibly a byte-swapped version of the same number on another machine.  This is followed by a 32 bit number in the same format which describes the number of bases in the file.   This is followed by the bases themselves packed two bases to the byte.  The first base is packed in the high order 4 bits (nibble), the second base in the low order four bits.  In C code:

byte = (base1<<4) + base2

The numerical values for the bases are:

0 – T,  1 – C,  2 – A,  3 – G,  4 – N (unknown)

The most significant bit in a nibble is set if the base is masked.


The gfServer program requires approximately 1 byte for every 3 bases in the genome it is indexing in DNA mode, and 1.5 bytes for each unmasked base in translated mode. The blat program requires approximately two bytes for each base in the genome in DNA mode, and three bytes for each base in translated mode. The other programs use relatively little memory.