April, 2005
Table of Contents
Is
T-Coffee different from ClustalW?
What
Can T-Coffee do for you (and what it cannot do)
Combining
Sequences and Structures
Identifying
Occurrences of a Motif: Mocca
Extended
Installation and other Packages
Default
Usage and Configuration Files
Library
Computation: Extension
Pairwise
Alignment Computation
Multiple
Alignment Computation
-multi_thread [NOT IMPLEMENTED]
Aligning
more than 100 sequences with DPA
Using
PDB templates for the Sequences
-domain_interactive [Examples]
USING NEW AND EXISTING METHODS
Methods
With An Established T-Coffee Interface
Methods
Without An Established T-Coffee Interface.
Abnormal
Terminations and Wrong Results
License and Terms of Use
Please make sure you have agreed with the terms
of the license attached to the package before using the T-Coffee package or its
documentation. T-Coffee is a freeware open source distributed under a GPL
license. This means that there is no restriction to its use, either in an
academic or a non academic environment.
Addresses and Contacts
We are always very eager to get some user
feedback. Please do not hesitate to drop us a line on:
cedric.notredame@europe.com
The latest updates of T-Coffee are always
available from:
igs-server.cnrs-mrs.fr/~cnotred
On this address you will also find a link to
some of the online T-Coffee servers, including Tcoffee@igs
igs-server.cnrs-mrs.fr/Tcoffee/
T-Coffee can be used to automatically check if
an updated version is available
t_coffee -update
Citations
It is
important that you cite T-Coffee when you use it. Citing us is (almost) like
giving us money: it helps us convincing our institutions that what we do is
useful and that they should keep paying our salaries and delivering Donuts to
our offices from time to time (Not that they ever did it, but it would be nice
anyway).
Cite the
server if you used it, otherwise, cite the original paper from 2000 (No, it was
never named "T-Coffee 2000").
|
1: |
||
|
|
T-Coffee:
A novel method for fast and accurate multiple sequence alignment. |
|
Other
useful publications include:
T-Coffee
|
1:
|
||
|
|
CaspR: a
web server for automated molecular replacement using homology modelling. |
|
|
2:
|
||
|
|
3DCoffee@igs:
a web server for combining sequences and structures into a multiple sequence
alignment. |
|
|
3:
|
||
|
|
3DCoffee:
combining protein sequences and structures within multiple sequence
alignments. |
|
|
4:
|
||
|
|
Tcoffee@igs:
A web server for computing, evaluating and combining multiple sequence
alignments. |
|
|
5:
|
||
|
|
Mocca:
semi-automatic method for domain hunting. |
|
|
6:
|
||
|
|
T-Coffee:
A novel method for fast and accurate multiple sequence alignment. |
|
|
7:
|
||
|
|
COFFEE:
an objective function for multiple sequence alignments. |
|
Mocca
|
8:
|
||
|
|
Mocca:
semi-automatic method for domain hunting. |
|
CORE
http://igs-server.cnrs-mrs.fr/~cnotred/Publications/Pdf/core.pp.pdf
DESCRIPTION
What is T-Coffee?
Before
going deep into the core of the matter, here are a few words to quickly explain
some of the things T-Coffee will do for you.
What does it do?
T-Coffee is
a multiple sequence alignment program: given a set of sequences previously
gathered using database search programs like BLAST, FASTA or Smith and
Waterman, T-Coffee will produce a multiple sequence alignment. To use T-Coffee
you must already have your sequences.
What can it align?
T-Coffee
will align DNA and protein sequences alike. It will be able to use structural
information for protein sequences with a known structure. We recently
introduced a new mode that makes T-Coffee able to accurately align large
datasets.
How can I use it?
T-Coffee is
not an interactive program. It runs from your UNIX or Linux command line and
you must provide it with the correct parameters. If you do not like typing
commands, here is the simplest available mode where T-Coffee only needs the
name of the sequence file:
t_coffee sample_seq1.fasta [**]
If
installing and running T-Coffee locally is beyond your computer abilities, we
propose you use one of the available online servers.
Is T-Coffee different from ClustalW?
According
to several benchmarks, T-Coffee appears to be more accurate than ClustalW. Yet,
this increased accuracy comes at a price: T-Coffee is slower than Clustal
(about N times).
If you are
familiar with ClustalW, or if you run a ClustalW server, you will find that we
made some efforts to ensure as much compatibility as possible between ClustalW
and T-COFFEE. Whenever it was relevant, we have kept the flag name and the flag
syntax of ClustalW. Yet, you will find that T-Coffee also has many extra
possibilities…
If you want
to align closely related sequences, T-Coffee can also be used in a fast mode,
much faster than ClustalW, and about as accurate ( T-Coffee -very_fast) This
mode is especially useful to align long sequences.
What Can T-Coffee do for you (and what it
cannot do)
(NOT) Fetching Sequences
T-Coffee
will NOT fetch sequences for you: you must select the sequences you want
to align before hand. We suggest you use any BLAST server and format your
sequences in FASTA so that T-COFFEE can use them easily.
Aligning Sequences
Making
accurate multiple alignments of DNA, RNA or Protein sequences.
Combining Alignments
T-Coffee
allows you to combine results obtained with several alignment methods. For
instance if you have an alignment coming from ClustalW, an other alignment
coming from Dialign, and a structural alignment of some of your sequences,
T-Coffee will combine all that information and produce a new multiple sequence
alignment having the best agreement with all these methods (see the FAQ for
more details)
t_coffee –in= Asample_aln1.aln,Asample_aln2.aln,Asample_aln3.aln
–outfile=combined_aln.aln [**]
Evaluating Alignments
You can use
T-Coffee to measure the reliability of your Multiple Sequence alignment. If you
want to find out about that, read the FAQ or the documentation for the -output
flag.
t_coffee
–infile=sample_aln1.aln –special_mode=evaluate [**]
Combining Sequences and Structures
One of the
latest improvements of T-Coffee is to let you combine sequences and structures,
so that your alignments are of higher quality. You need to have sap package
installed to fully benefit of this facility.
t_coffee 3d.fasta –special_mode=3dcoffee
[*#]
Using this
mode will cause T-Coffee to automatically identify the target corresponding to
your sequence as indicated by an NCBI BLAST. T-Coffee then obtains the required
PDB sequences from RCSB. However, if you are also using –template_file, the
program will use the template you specified and the corresponding files on your
disk.
All these
network based operations are carried out using wget. If wget is not installed
on your system, you can get it for free from (www.wget.org). To make sure wget
is installed on your system, type
which wget [**]
Identifying Occurrences of a Motif:
Mocca
Mocca is a
special mode of T-Coffee that allows you to extract a series of repeats from a
single sequence or a set of sequences. In other words, if you know the
coordinates of one copy of a repeat, you can extract all the other occurrences.
If you want to use Mocca, simply type:
t_coffee –other_pg mocca
sample_seq1.fasta [**]
The program
needs some time to compute a library and it will then prompt you with an
interactive menu. Follow the instructions.
How Does T-Coffee works
If you only
want to make a standard multiple alignments, you may skip these explanations.
But if you want to do more sophisticated things, these few indications may help
before you start reading the doc and the papers.
When you run
T-Coffee, the first thing it does is to compute a library. The library is a
list of pairs of residues that could be aligned. It is like a Xmas list:
you can ask anything you fancy, but it is down to Santa to assemble a
collection of Toys that won't get him stuck at the airport, while going through
the metal detector.
Given a
standard library, it is not possible to have all the residues aligned at the
same time because all the lines of the library may not agree. For instance,
line 1 may say
Residue 1 of
seq A with Residue 5 of seq B,
and line
100 may say
Residue 1 of
seq A with Residue 29 of seq B,
Each of
these constraints comes with a weight and in the end, the T-Coffee algorithm
tries to generate the multiple alignment that contains constraints whose sum of
weights yields the highest score. In other words, it tries to make happy as
constraints as possible (replace the word constraint with, friends, family
members, collaborators… and you will know exactly what we mean).
You can
generate this list of constraints however you like. You may even provide it
yourself, forcing important residues to be aligned by giving them high weights
(see the FAQ). For your convenience, T-Coffee can generate (this is the
default) its own list by making all the possible global pairwise alignments,
and the 10 best local alignments associated with each pair of sequences. Each
pair of residues observed aligned in these pairwise alignments becomes a line
in the library.
Yet be aware
that nothing forces you to use this library and that you could build it using
other methods (see the FAQ). In protein language, T-COFEE is synonymous for
freedom, the freedom of being aligned however you fancy (It is with that sort
of statements that I got elected Chief Tryptophan Officer in some previous
life).
INSTALLATION
Standard Installation
1-decompress distribution.tar.gz
gunzip
distribution.tar.gz
2-untar distribution.tar
tar -xvf
distribution.tar
3-This will create
the distribution directory with the following structure:
distribution/bin
distribution/doc/t_coffee_doc.pdf,t_coffee_doc.html
distribution/t_coffee_source
distribution/example
distribution/html
4-go into the main directory and type:
./install
You will
know the installation proceeded completely if the mention:
Installation
of t_coffee Successful
appears on
your screen to indicate a proper completion.
5-add the bin folder
to your path:
set path
= ($path . <address of the t_coffee bin folder>)
Note: The latest t_coffee distribution (2.15 and higher) is self
contained and only requires one executable. You may still require external
modules (sap, blast, ClustalW) if you wish to use another mode than the
default.
Note: When updating, make sure to remove the old distribution and any
associated program from your path.
6-If you have PDB installed:
Assuming you
have a standard PDB installation in your file system
setenv
PDB_DIR pdb_dir/structures/all/
Note: This must be added to your login file.
Extended Installation and other Packages
By default,
T-Coffee does not require any other package than those included in the
distribution. However, depending on your needs, you may want to install some of
the following:
Package Function
===========================================================
ClustalW can interact with t_coffee
-----------------------------------------------------------
wget 3DCoffee
Automatic Downloading of
Structures
Remote
use of the Fugue server
-----------------------------------------------------------
sap structure/structure
comparisons (obtain it from
-----------------------------------------------------------
Once the
package is installed, make sure make sure that the executable is on your path,
so that t_coffee can find it automatically.
QUICK START
T-COFFEE
Write your
sequences in the same file (Swiss-prot, Fasta or Pir) and type.
t_coffee sample_seq1.fasta
[**]
This will
output two files:
sample_seq1.aln
A multiple
alignment in a format similar to ClustalW, that can be read by most programs.
sample_seq1.dnd
A
dendrogram in newick format
Note: Using the dendrogram in place of a phylogenetic tree is something
very bad, that will land you straight to hell, or in jail, or any place you
don’t like, or turn . If you need a tree, compute one using your MSA and an
appropriated phylogenetic package like phylip
MOCCA
Write your
sequences in the same file (Swiss-prot, Fasta or Pir) and type.
t_coffee –other_pg mocca
sample_seq1.fasta [**]
This command output one files (<your sequences>.mocca_lib) and
starts an interactive menu.
Note: The computation of the mocca_lib file can take some time. Whenever
you re-run mocca on the sane sequences, it will start looking for the mocca_lib
file and read it, or will generate it if it does not exist.
RECENT
MODIFICATIONS
A detailed log
of the modifications (with versions), can be found in the to_do.txt file
in the doc/ folder.
The most
notable modifications have to do with the structure of the input. From version
2.20, all files must be tagged to indicate their nature (A: alignment, S:
Sequence, L: Library…). We are becoming stricter, but that’s for your own good…
MANUAL
This manual is only at a very preliminary stage
of redaction. It will only show you how to do the very basic with T-Coffee. In
order to solve a more specific problem, or answer a query, we suggest you first
go through the FAQ to see of your problem has been addressed, read it and then
read carefully the documentation associated with corresponding flags. Of
course, we also welcome queries and do our best to provide answers and clues in
a timely manner.
Using T-Coffee
Standard Alignments
T-Coffee can align sequences, structures and
profiles. The default mode when using t_coffee is:
t_coffee sample_seq1.fasta
It is also possible to combine sequences from
various sources:
t_coffee sample_seq1.fasta sample_seq2.fasta
Or even, sequences coming from sequences and
alignment files:
t_coffee sample_seq1.fasta,Ssample_aln2.aln
Note the ‘S’ identifier, that tells the program to
use the alignment as a collection of unaligned sequences.
Alignment Combination
It is possible to combine several alignments
into one final alignment,
t_coffee –in Asample_aln1_1.aln,Asample_aln1_2.aln
–outfile=combined_aln.aln
Note the ‘A’ identifier that tells the program to
keep the sequences aligned.
Aligning Sequences and Structures
Assuming some structures are associated with
your sequences, it is possible to align these sequences while using associated
structural information. The easiest way to do this is to use 3dcoffee:
t_coffee 3d_coffee
Aligning Sequences and Profiles
T-Coffee can make multiple profile alignments.
In this context, the alignments are treated as single sequences and aligned to
one another in a progressive fashion. Currently, we only support profiles under
the form of standard multiple sequence alignments. The profile must either be
entered via the –profile flag:
t_coffee –profile sample_aln1.aln,sample_aln2.aln –outfile=combined_profiles.aln
It is also possible to read the profile via the
–in flag, as long as they are preceded with the ‘R’ identifier.
t_coffee Ssample_seq1.fasta,Rsample_aln2.aln
–outfile seqprofile_aln
All the internal methods should support profiles.
External methods do not support profiles (unless specified otherwise).
Using Structures (Or templates) Within
Profiles
If your profiles contain structures, you can make
sure these will be used during the computatiuon by specifying the 3dcoffee
special mode:
t_coffee Rsample_profile1.aln,Rsample_profile2.aln
–special_mode=3dcoffee –outfile=aligned_prf.aln
Note that when providing a collection of
templates, the program will use the –template_file flag to look for templates
within the sequences AND within the profiles associated with some sequences.
Using New and Existing Methods
Although,
it does not necessarily do so explicitly, T-Coffee always end up combining
libraries. Libraries are collections of pairs of residues. Given a set of
libraries, T-Coffee makes an attempt to assemble the alignment with the highest
level of consistence. You can think of the alignment as a timetable. Each
library pair would be a request from students or teachers, and the job of
T-Coffee would be to assemble the time table that makes as many people as
possible happy…
In
T-Coffee, methods replace the students/professors as constraints generators.
These methods can be any standard/non standard alignment methods that can be
used to generate alignments (pairwise, most of the time). These alignments can
be viewed as collections of constraints that must be fit within the final
alignment. Of course, the constraints do not have to agree with one another…
This
section shows you what are the vailable method in T-Coffee, and how you can add
your own methods, either through direct parameterization or via a perl script.
Using Methods Integrated in T-Coffee
Some packages
already have an interface with t_coffee, these include:
align_pdb: ALIGN_PDB_4_TCOFFEE
sap: SAP_4_TCOFFEE
lalign2list: LALIGN_4_TCOFFEE
clustalw: CLUSTALW_4_TCOFFEE
If these
programs are installed on your system and you want t_coffee to use a specific
version:
setenv
CLUSTALW_4_TCOFFEE <path to your version>
Built in
methods methods can be requested using the following names:
fast_pair
Makes a
global fasta style pairwise alignment. For proteins, matrix=blosum62mt, gep=-1,
gop=-10, ktup=2. For DNA, matrix=idmat (id=10), gep=-1, gop=-20, ktup=5. Each
pair of residue is given a score function of the weighting mode defined by
-weight.
slow_pair
Identical
to fast pair, but does a full dynamic programming, using the myers and miller
algorithm. This method is recommended if your sequences are distantly related.
ifast_pair [unsupported]
Makes a
global fasta alignmnet using the previously computed pairs as a library. `i`
stands for iterative. Each pair of residue is given a score function of the
weighting mode defined by -weight.
align_pdb_pair
[unsupported]
Uses the
align_pdb routine to align two structures. The pairwise scores are those
returnes by the align_pdb program. If a structure is missing, fast_pair is used
instead. Each pair of residue is given a score function defined by align_pdb.
sap_pair
Uses sap to
align two structures. Each pair of residue is given a score function defined by
sap. You must have sap installed on your system to use this method.
clustalw_pair
Uses
clustalw (default parameters) to align two sequences. Each pair of residue is
given a score function of the weighting mode defined by -weight.
clustalw_aln
Makes a
multiple alignment using ClustalW and adds it to the library. Each pair of
residue is given a score function of the weighting mode defined by -weight.
lalign_id_pair
Same as
lalign_rs_pir, but using the level of identity as a weight.
lalign_id_m_pair
Same as
above, but does the alignment both way (m stands for miror).
lalign_s_pair
Same as
above but does also the self comparison (s stands for self). This is needed
when extracting repeats. The weights used that way are based on identity.
lalign_rs_s_pair
Same as
above but does also the self comparison (s stands for self). This is needed
when extracting repeats.
matrix
Amy matrix
can be requested. Simply indicate as a method the name of the matrix preceded
with an X (i.e. Xpam250mt). If you indicate such a matrix, all the other
methods will simply be ignored, and a standard fast progressive alignment will
be computed. If you want to change the substitution matrix used by the methods,
use the –matrix flag.
fast_cdna_pair
[unsupported]
This method
computes the pairwise alignment of two cDNA sequences. It is a fast_pair
alignment that only takes into account the amino-acid similarity and uses
different penalties for amino-acid insertions and frameshifts.
This
alignment is turned into a library where matched nucleotides receive a score
equql to the average level of identity at the amino-acid level.
This mode
is intended to clean cDNA obtained from ESTs, or to align pseudo-genes.
To request a
method, see the -in flag. For instance, if you wish to request the use of
fast_pair and lalign_id_pair (the current default):
t_coffee -in
Ssample_seq1.fasta,Mfast_pair,Mlalign_id_pair [**]
The order in
which methods are fed is irrelevant.
Integrating External Methods
Direct access to external methods
A special method exists in T-Coffee that can be
used to invoke any existing program:
t_coffee sample_seq1.fasta
–in=Mem0clustalw0pairwise [**]
In this context, Clustalw is a method that can
be ran with the following command line:
method
–infile=<infile> -outfile=<outfile>
Clustalw can be replaced with any method using
a similar syntax. If the program you want to use cannot be run this way, you
can either write a perl wrapper that fits the bill or write a tc_method file
adapted to your program (cf next section).
This special method (em, external method) uses
the following syntax:
Em0<method>0<aln_mode:pairwise¦
s_pairwise|multiple>
Note: The 0 is used as a separator.
This symbol must not be part of the method name.
Customizing an external method (with
parameters) for T-Coffee
T-Coffee can run external methods, using a tc_method
file that can be used in place of an established method. Two such files are
incorporated in T-Coffee. You can dump them and customize them according to
your needs:
For instance if you have ClustalW installed,
you can use the following file to run the method on your dataset:
t_coffee –other_pg
unpack_clustalw_method.tc_method [**]
t_coffee –other_pg unpack_generic_method.tc_method
[**]
The second file (generic_method.tc_method)
contains many hints on how to customize your new method. The first file is a
very straightforward example on how to have t_coffee to run Clustalw with a set
of parameters you may be interested in:
*TC_METHOD_FORMAT_01
***************clustalw_method.tc_method*********
EXECUTABLE clustalw
ALN_MODE pairwise
IN_FLAG -INFILE=
OUT_FLAG -OUTFILE=
OUT_MODE aln
PARAM -gapopen=-10
SEQ_TYPE S
*************************************************
This configuration file will cause T-Coffee to emit the following system call:
clustalw –INFILE=tmpfile1 –OUTFILE=tmpfile2 –gapopen=-10
Note that ALN_MODE instructs t_coffee to run clustalw on every pair of sequences (cf generic_method.tc_method for more details).
The tc_method files are treated like any standard established method in T-Coffee. For instance, if the file clustalw_method.tc_method is in your current directory, run:
t_coffee sample_seq1.fasta
–in=Mclustalw_method.tc_method [**]
Managing a collection of tc_method files
It may be convenient to store all the method files in a single location on your system. By default, t_coffee will go looking into the directory ~/.t_coffee/methods/. You can change this by either:
-Modifying the METHODS_4_TCOFFEE in define_headers.h
-recompile
OR
setenv METHODS_4_TCOFFEE <another location>
Advanced Method Integration
It may sometimes be difficult to customize the program you want to use through a tc_method file. In that case, you may rather use an external perl_script to run your external application. This can easily be achieved using the generic_method.tc_method file.
*TC_METHOD_FORMAT_01
***************generic_method.tc_method*********
EXECUTABLE tc_generic_method.pl
ALN_MODE pairwise
IN_FLAG -infile=
OUT_FLAG -outfile=
OUT_MODE aln
PARAM -method clustalw
PARAM -gapopen=-10
SEQ_TYPE S
*************************************************
* Note: &bsnp can be used to for white spaces
When you run this method:
t_coffee sample_seq1.fasta
–in=Mgeneric_method.tc_method [**]
T-Coffee runs the script tc_generic_method.pl on your data. It also provides the script with parameters. In this case –method clustalw indicates that the script should run clustalw on your data. The script tc_generic_method.pl is incorporated in t_coffee. Over the time, this script will be the place where novel methods will be integrated
will be used to run the script tc_generic_method.pl. The file tc_generic_method.pl is a perl file, automatically generated by t_coffee. Over the time this file will make it possible to run all available methods.
You can dump the script using the following command:
t_coffee –other_pg=unpack_tc_generic_method.pl
[**]
Note: If there is a copy of that
script in your local directory, that copy will be used in place of the internal
copy of T-Coffee.
Reference of tc_method file
*TC_METHOD_FORMAT_01
******************generic_method.tc_method*************
*
* Incorporating new methods in T-Coffee
* Cedric Notredame 17/04/05
*
*******************************************************
*This file is a method file
*Copy it and adapt it to your need so that the method
*you want to use can be incorporated within T-Coffee
*******************************************************
* USAGE *
*******************************************************
*This file is passed to t_coffee via –in:
*
* t_coffee –in Mgeneric_method.method
*
* The method is passed to the shell using the following
*call:
*<EXECUTABLE><IN_FLAG><seq_file><OUT_FLAG><outname><PARAM>
*
*Conventions:
*<FLAG_NAME> <TYPE> <VALUE>
*<VALUE>: no_name <=> Replaced with a space
*<VALUE>:   <=> Replaced with a space
*
*******************************************************
* EXECUTABLE *
*******************************************************
*name of the executable
*passed to the shell: executable
*
EXECUTABLE tc_generic_method.pl
*
*******************************************************
* ALN_MODE *
*******************************************************
*pairwise ->all Vs all (no self )[(n2-n)/2aln]
*m_pairwise ->all Vs all (no self)[n^2-n]^2
*s_pairwise ->all Vs all (self): [n^2-n]/2 + n
*multiple ->All the sequences in one go
*
ALN_MODE pairwise
*
*******************************************************
* OUT_MODE *
*******************************************************
* mode for the output:
*External methods:
* aln -> alignmnent File (Fasta or ClustalW Format)
* list-> List file (TC_LIB_FORMAT_01)
*Internal Methods:
* fL -> Internal Function returning a List (Librairie)
* fA -> Internal Function returning an Alignmnent
*
OUT_MODE aln
*
*******************************************************
* IN_FLAG *
*******************************************************
*IN_FLAG
*flag indicating the name of the in coming sequences
*IN_FLAG S no_name ->no flag
*IN_FLAG S  –in  -> “ –in “
*
IN_FLAG -infile=
*
*******************************************************
* OUT_FLAG *
*******************************************************
*OUT_FLAG
*flag indicating the name of the out-coming data
*same conventions as IN_FLAG
*OUT_FLAG S no_name ->no flag
*
OUT_FLAG -outfile=
*
*******************************************************
* SEQ_TYPE *
*******************************************************
*G: Genomic, S: Sequence, P: PDB, R: Profile
*Examples:
*SEQTYPE S sequences against sequences (default)
*SEQTYPE S_P sequence against structure
*SEQTYPE P_P structure against structure
*SEQTYPE PS mix of sequences and structure
*
SEQ_TYPE S
*
*******************************************************
* PARAM *
*******************************************************
*Parameters sent to the EXECUTABLE
*If there is more than 1 PARAM line, the lines are
*concatenated
*
PARAM -method clustalw
PARAM -OUTORDER=INPUT -NEWTREE=core -align -gapopen=-15
*
*******************************************************
* END *
*******************************************************
Creating Your Own T-Coffee Libraries
If the
method you want to use is not integrated, or impossible to integrate, you can
generate your own libraries, either directly or by turning existing alignments
into libraries.
Using Pre-Computed Alignments
If the method
you wish to use is not supported, or if you simply have the alignments, the
simplest thing to do is to generate yourself the pairwise/multiple alignments,
in FASTA, ClustalW, msf or Pir format and feed them into t_coffee using the -in flag:
t_coffee
–in=Asample_aln1_1.aln,Asample_aln1_2.aln –outfile=combined_aln.aln [**]
Customizing the Weighting Scheme
The previous
integration method forces you to use the same weighting scheme for each
alignment and the rest of the libraries generated on the fly. This weighting
scheme is based on global pairwise sequence identity. If you want to use a more
specific weighting scheme with a given method, you should either
-generate your own library (cf next
section)
-convert your alignments into a library,
using the –weight flag:
t_coffee –in Asample_aln1.aln
–out_lib=test_lib.tc_lib –lib_only –weight=sim_pam250mt [**]
t_coffee
–in=Asample_aln1.aln,Ltest_lib.tc_lib –outfile=outaln [**]
Note: Default methods are reset when you explicitly use -in, if you wish
to keep using fast_pair and lalign_id_pair, you need to indicate these methods
explicitly:
t_coffee
–in=Asample_aln1_1.aln,Asample_aln1_2.aln,Mfast_pair,Mlalign_id_pair
–outfile=out_aln [**]
Generating Your Own Libraries
This is
suitable if you have local alignments, or very detailed information about your
potential residue pairs, or if you want to use a very specific weighting
scheme. You will need to generate your own libraries, using the format
described in the last section.
Note: You can have up to 200 libraries. They do not need to contain the
same sequences.
The
Notion of Templates in T-Coffee [TO DO]
FAQ
Abnormal Terminations and Wrong Results
In order to
maintain t_coffee and fix bugs and problem, we need to get as much fee
Q: The program keeps crashing when I give my sequences
A: This may be
a format problem. Try to reformat your sequences using any utility
(readseq...). We recommend the Fasta format. If the problem persists, contact
us.
A: Your
sequences may not be recognized for what they really are. Normally T-Coffee
recognize the type of your sequences automatically, but if it fails, use:
t_coffee
sample_seq1.fasta -type=PROTEIN [**]
Q: The default alignment is not good enough
A: see next
question
Q: The alignment contains obvious mistakes
A: This
happens with most multiple alignment procedures. However, wrong alignments are
sometimes caused by a bugs or an implementation mistake. Please report the most
unexpected results to the authors.
Q: The program is crashing
A: If you get
the message:
FAILED TO
ALLOCATE REQUIRED MEMORY
See the next
question.
If the program
crashes for some other reason, please check whether you are using the right
syntax and if the problem persists get in touch with the authors.
Q: I am running out of memory
A: You can use
a more accurate, slower and less memory hungry dynamic programming mode called
myers_miller_pair_wise. Simply indicate the flag:
t_coffee
sample_seq1.fasta –special_mode low_memory [**]
Note that this
mode will be much less time efficient than the default, although it may be
slightly more accurate. In practice the parameterization associate with special
mode turns off every memeory expensive heuristic within T-Coffee. For version
2.11 this amounts to
t_coffee sample_seq1.fasta
-in=Mslow_pair,Mlalign_id_pair -tree_mode=slow -dp_mode=myers_miller_pair_wise
[**]
If you keep running out of memory, you may also
want to lower –maxnseq, to ensure that t_coffee_dpa will be used.
Input/Output Control
Q: How many Sequences can t_coffee handle
A: T-Coffee is
limited to a maximum of 50 sequences. Above this number, the program
automatically switches to a heuristic mode, named DPA, where DPA stands for
Double Progressive Alignment.
DPA is still
in development and the version currently shipped with T-Coffee is only a beta
version.
Q: How many ways to pass parameters to t_coffee?
A: See the
section well behaved parameters
Q: How can I change the default output format?
A: See the -output option, common output formats
are:
t_coffee
sample_seq1.fasta -output=msf,fasta_aln [**]
Q: My sequences are slightly different between all the alignments.
A: It does not
matter. T-Coffee will reconstruct a set of sequences that incorporates all the
residues potentially missing in some of the sequences ( see flag -in).
Q: Is it possible to pipe stuff OUT of t_coffee?
A: Specify
stderr or stdout as output filename, the output will be redirected accordingly.
For instance
t_coffee
sample_seq1.fasta -outfile=stdout -out_lib=stdout [**]
This
instruction will output the tree (in
Q: Is it possible to pipe stuff INTO t_coffee?
A: If as a
file name, you specify stdin, the content of this file will be expected
throught pipe:
cat
sample_seq1.fasta | t_coffee -infile=stdin [**]
will be
equivalent to
t_coffee sample_seq1.fasta
[**]
If you do not
give any argument to t_coffee, they will be expected to come from pipe:
cat sample_param_file.param |
t_coffee -parameters=stdin [**]
For instance:
echo
–in=Ssample_seq1.fasta,Mclustalw_pair | t_coffee –parameters=stdin [**]
Q:
Can I read my parameters from a file?
A: See the well behaved parameters section.
Q: I want to decide myself on the
name of the output files!!!
A: Use the -run_name flag.
t_coffee
sample_seq1.fasta –run_name=guacamole [**]
Q: I want to use the sequences in an alignment file
A: Simply fed
your alignment, any way you like, but do not forget to append the prefix S for
sequence:
t_coffee
Ssample_aln1.aln [**]
t_coffee
-infile=Ssample_aln1.aln [**]
t_coffee
–in=Ssample_aln1.aln,Mslow_pair,Mlalign_id_pair –outfile=outaln
This means that the gaps will be reset and that
the alignment you provide will not be considered as an alignment, but as a set
of sequences.
Q: I only want to produce a library
A: use the
–lib_only flag
t_coffee
sample_seq1.fasta -out_lib=sample_lib1.tc_lib -lib_only [**]
Please, note
that the previous usage supersedes the use of the –convert flag. Its main
advantage is to restrict computation time to the actual library computation.
Q: I want to turn an alignment into a library
A: use the
–lib_only flag
t_coffee
–in=Asample_aln1.aln -out_lib=sample_lib1.tc_lib -lib_only [**]
It is also
possible to control the weight associated with this alignment (see the –weight section).
t_coffee –in=Asample_aln1.aln
-out_lib=sample_lib1.tc_lib -lib_only –weight=1000 [**]
Q: I want to concatenate two libraries
A: You cannot
concatenate these files on their own. You will have to use t_coffee. Assume you
want to combine tc_lib1.tc_lib and tc_lib2.tc_lib.
t_coffee -in
Lsample_lib1.tc_lib Lsample_lib2.tc_lib –lib_only -out_lib=sample_lib3.tc_lib
[**]
Q: What happens to the gaps when an alignment is fed to T-Coffee
A: An
alignment is ALWAYS considered as a
library AND a set of sequences. If you want your alignment to be considered as
a library only, use the S identifier.
t_coffee
Ssample_aln1.aln –outfile=outaln[**]
It will be seen as a sequence file, even if it
has an alignment format (gaps will be removed).
Q: I cannot print the html graphic display!!!
A: This is a
problem that has to do with your browser. Instead of requesting the score_html
output, request the score_ps output that can be read using ghostview:
t_coffee
sample_seq1.fasta -output=score_ps [**]
or
t_coffee
sample_seq2.fasta -output=score_pdf [**]
Note: you need to have the converter ps2pdf installed on your system
(standard under Linux and cygwin).
Note: the latest versions of Internet Explorer and Netscape now allow
the user to print the HTML display. Do
not forget to request Background printing.
Q: I want to output an html file and a regular file
A: see the
next question
Q: I would like to output more than one alignment format at the same
time
A: The flag -output accepts more than one
parameter. For instance,
t_coffee
sample_seq1.fasta -output=clustalw,score_html,score_ps,msf [**]
This will output four alignment files in the
corresponding formats. Alignments' names will have the format name as an
extension.
Alignment Computation
Q: I do not want to compute the alignment.
A: use the
-convert flag
t_coffee
sample_aln1.aln -convert -output=gcg [**]
This command will
read the .aln file and turn it into an .msf alignment.
Q: I would like to force some residues to be aligned.
If you want to
brutally force some residues to be aligned, you may use as a post processing,
the force_aln function of seq_reformat:
t_coffee –other_pg
seq_reformat –in sample_aln4.aln –action +force_aln seq1 10 seq2 15 [**]
t_coffee –other_pg
seq_reformat –in sample_aln4.aln –action +force_aln sample_lib4.tc_lib02 [**]
sample_lib4.tc_lib02
is a T-Coffee library using the tc_lib02 format:
*TC_LIB_FORMAT_02
SeqX resY
ResY_index SeqZ ResW ResW_index
…
Warning: the TC_LIB_FORMAT_02 is still experimental and unsupported. It can only be used in the context of the force_aln function described here.
Given more
than one constraint, these will be applied one after the other, in the order
they are provided. This greedy procedure means that the Nth constraint may
disrupt the (N-1)th previously imposed constraint, hence the importance of
forcing the constraints in the right order, with the most important coming
last.
We do not
recommend imposing hard constraints on an alignment, and it is much more
advisable to use the soft constraints provided by standard t_coffee libraries
(cf. building your own libraries section)
Q: I would like to use structural alignments.
See the
section Using structures in Multiple Sequence Alignments, or see the question I want to build my own libraries.
Q: I want to build my own libraries.
A: Turn your
alignment into a library, forcing the residues to have a very good weight,
using structure:
t_coffee –in
Asample_seq1.aln -weight=1000 -out_lib=sample_seq1.tc_lib –lib_only [**]
The value 1000
is simply a high value that should make it more likely for the substitution
found in your alignment to reoccur in the final alignment. This will produce
the library sample_aln1.tc_lib that you can later use when aligning all the
sequences:
t_coffee –in
Ssample_seq1.fasta Lsample_seq1.tc_lib –outfile sample_seq1.aln [**]
If you only
want some of these residues to be aligned, or want to give them individual
weights, you will have to edit the library file yourself or use the –force_aln
option (cf FAQ: I would like to force some residues to be aligned). A value of
N*N * 1000 (N being the number of sequences) usually ensure the respect of a
constraint.
Q: I want to use my own tree!!!!
A: Use the -usetree=<your own tree> flag.
t_coffee
sample_seq1.fasta –usetree=sample_tree.dnd [**]
Q: I
want to align coding DNA
A: use the
fasta_cdna_pair method that compares two cDNA using the best reading frame and
taking frameshifts into account.
Note: This method has not yet been fully tested and is only provided “as-is”
with no warranty.
t_coffee
sample_seq4.fasta –in Mcdna_fast_pair [**]
Notice that in
the resulting alignments, all the gaps are of modulo3, except one small gap in
the first line of sequence hmgl_trybr. This is a framshift, made on purpose.
You can realign the same sequences while ignoring their coding potential and
treating them like standard DNA:
t_coffee
sample_seq4.fasta [**]
Q: I do not want to use all the possible pairs when computing the
library
Q: I only want to use specific pairs to compute the library
A: Simply
write in a file the list of sequence groups you want to use:
t_coffee
sample_seq1.fasta –in=Mclustalw_pair,Mclustalw_aln
–lib_list=sample_list1.lib_list –outfile=test
***************sample_list1.lib_list************
2 hmgl_trybr hmgt_mouse
2 hmgl_trybr hmgb_chite
2 hmgl_trybr hmgl_wheat
3 hmgl_trybr hmgl_wheat hmgl_mouse
***************sample_list1.lib_list************
Note: Pairwise methods (slow_pair…) will only be applied to list of
pairs of sequences, while multiple methods (clustalw_aln) will be applied to
any dataset having more than two sequences.
Q:
There are duplicates or quasi-duplicates in my set
A: If you can
remove them, this will make the program run faster, otherwise, the t_coffee
scoring scheme should be able to avoid over-weighting of over-represented
sequences.
Using Structures and Profiles
Q: Can I align sequences to a
profile with T-Coffee ?
A: Yes, you simply need to indicate that your
alignment is a profile with the R tag..
t_coffee sample_seq1.fasta Rsample_aln2.aln
–outfile tacos
Q: Can I align sequences Two or More
Profiles ?
A: Yes, you, simply tag your profiles with the
letter R and the program will treat them like standard sequences.
t_coffee Rsample_aln1.fasta Rsample_aln2.aln
–outfile tacos
Q: Can I align two profiles
according to the structures they contain?
A: Yes. As long as the structure sequences are
named according to their PDB identifier
t_coffee Rsample_profile1.aln,Rsample_profile2.aln
–special_mode=3dcoffee –outfile=aligne_prf.aln
Alignment Evaluation
Q: How good is my alignment?
A: see what is the color index?
Q: What is that color index?
A: T-Coffee
can provide you with a measure of consistency among all the methods used. You
can produce such an output using:
t_coffee
sample_seq1.fasta -output=score_html [**]
This will
compute your_seq.score_html that you can view using netscape. An alternative is
to use score_ps or score_pdf that can be viewed using ghostview or acroread,
score_ascii will give you an alignment that can be parsed as a text file.
A book chapter
describing the CORE index is available on:
http://igs-server.cnrs-mrs.fr/~cnotred/Publications/Pdf/core.pp.pdf
Q: Can I evaluate alignments NOT
produced with T-Coffee?
A: Yes. You
may have an alignment produced from any source you like. To evaluate it do:
t_coffee
–infile=sample_aln1.aln - in=Lsample_aln1.tc_lib –special_mode=evaluate [**]
If you have no
library available, the library will be computed on the fly using the following
command. This can take some time, depending on your sample size. To monitor the
progress in a situation where the default library is being built, use:
t_coffee
–infile=sample_aln1.aln –special_mode evaluate [**]
Q: Can I Compare Two Alignments?
A: Yes. You
can treat one of your alignments as a library and compare it with the second
alignment:
t_coffee
–infile=sample_aln1_1.aln -in=Asample_aln1_2.aln –special_mode=evaluate [**]
If you have no
library available, the library will be computed on the fly using the following
command. This can take some time, depending on your sample size. To monitor the
progress in a situation where the default library is being built, use:
t_coffee
–infile=sample_aln1.aln –special_mode evaluate [**]
Q: I am aligning sequences with long regions of very good overlapp
A: Increase
the ktuple size ( up to 4 or 5 for DNA) and up to 3 for proteins.
t_coffee
sample_seq1.fasta -ktuple=3
This will
speed up the program. It can be very useful, especially when aligning ESTs.
Q: Why is T-Coffee changing the names of my sequences!!!!
A: If there is
no duplicated name in your sequence set, T-Coffee's handling of names is
consistent with Clustalw, see Sequence Name Handling in the Format section.
If your
dataset contains sequences with identical names, these will automatically be
renamed to:
************************
>seq1
>seq1
************************
>seq1
>seq1_1
************************
The situation where this renaming creates
two sequence with a similar name is not currently supported.
FLAG
DESCRIPTION: REFERENCE
This
reference manual gives a list of all the flags that can be used to modify the
behavior of T-Coffee. For your convenience, we have grouped them according to
their nature. To display a list of all the flags used in the version of
T-Coffee you are using (along with their default value), type:
t_coffee [**]
Or
t_coffee –help [**]
Or
t_coffee –help –in
Or any
other parameter
Well Behaved Parameters
Separation
You can use
any kind of separator you want (i.e. ,; <space>=). The syntax used in
this document is meant to be consistent with that of ClustalW. However, in
order to take advantage of the automatic filename compleation provided by many
shells, you can replace “=” and “,” with a space.
Posix
T-Coffee is
not POSIX compliant.
Entering the right parameters
There are many
ways to enter parameters in T-Coffee, see the -parameter flag in
Parameters Syntax
Default Usage and Configuration
Files
In the
following documentation:
sample_seq.seq is provided in the
distribution, in the tutorial directory, along with all the other sample file
mentioned in this documentation. The file sample_seq.pep is assumed to be a
file containing sequences in any of the format recognized by T-Coffee.
If no
flag is used <your sequence> must be the first argument. See format for
further information.
t_coffee sample_seq1.fasta
[**]
Which
is equivalent to
t_coffee
Ssample_seq1.fasta [**]
When
you do so, sample_seq1 is used as a
name prefix for every file the program outputs.
Note: This is one of the exceptions (with –infile) where the identifier
tag (S,A,L,M…) can be omitted. Any dataset provided this way will be assumed to
be a sequence (S). These exceptions have been designed to keep the program
compatible with ClustalW.
Usage:
-parameters=parameters_file
Default:
no parameters file
Indicates
a file containing extra parameters. Parameters read this way behave as if they
had been added on the right end of the command line that they either
supersede(one value parameter) or complete (list of values). For instance, the
following file (parameter.file) could be used
*******sample_param_file.param***********
-in=Ssample_seq1.fasta,Mfast_pair
-output=msf_aln
*****************************************
Note: This parameter file can ONLY contain valid parameters. Comments
are not allowed. Parameters passed this way will be checked like normal
parameters.
Used
with:
t_coffee
-parameters=sample_param_file.param[**]
Will
cause t_coffee to apply the fast_pair method onto to the sequences contained in
sample_seq.fasta. If you wish, you can also pipe these arguments into t_coffee,
by naming the parameter file "stdin" (as a rule, any file named stdin
is expected to receive its content via the stdin)
cat
sample_param_file.param | t_coffee
-parameters=stdin [**]
-t_coffee_defaults
Usage:
-t_coffee_defaults=<file_name>
Default:
not used.
This
flag tells the program to use some default parameter file for t_coffee. The
format of that file is the same as the one used with -parameters. The file used
is either:
1. <file name> if a name has been
specified
2. ~/.t_coffee_defaults
if no file was specified
3. The file indicated by the environment
variable TCOFFEE_DEFAULTS
Usage:
-special_mode= hard coded mode
Default:
not used.
It
indicates that t_coffee will use some hard coded parameters. These include:
quickaln:
very fast approximate alignment
dali:
a mode used to combine dali pairwise alignments
evaluate:
defaults for evaluating an alignment
3dcoffee:
runs t_coffee with the 3dcoffee parameterisation
Other
modes exist that are not yet supported
-score [Deprecated]
Usage: -score
Default:
not used
Toggles
on the evaluate mode and causes t_coffee to evaluates a precomputed alignment
provided via -infile=<alignment>. The flag -output must be
set to an appropriate format (i.e. -output=score_ascii, score_html or
score_pdf). A better default parameterization is obtained when using the flag -special_mode=evaluate.
-evaluate
Usage:
-evaluate
Default:
not used
Replaces
–score. This flag toggles on the evaluate mode and causes t_coffee to evaluates
a pre-computed alignment provided via -infile=<alignment>. The
flag -output must be set to an appropriate format (i.e.
-output=score_ascii, score_html or score_pdf).
The
main purpose of –evaluate is to let you control every aspect of the evaluation.
Yet it is advisable to use pre-defined parameterization: special_mode=evaluate.
t_coffee
–infile=sample_aln1.aln -special_mode=evaluate [**]
t_coffee
–infile=sample_seq1.aln –in Lsample_lib1.tc_lib
–special_mode=evaluate [**]
Usage:
-convert
Default:
turned off
Toggles
on the conversion mode and causes T-Coffee to convert the sequences,
alignments, libraries or structures provided via the -infile and -in
flags. The output format must be set via the -output flag. This flag can
also be used if you simply want to compute a library (i.e. you have an
alignment and you want to turn it into a library).
This
flag is ClustalW compliant.
Usage: -do_align
Default:
turned on
For compatibility with
ClustalW
Special Parameters
Usage:
-version
Default:
not used
Returns
the current version number
Usage:
-check_configuration
Default:
not used
Checks
your system to determine whether all the programs T-Coffee can interact with
are installed.
Usage: -cache=<use, update, ignore,
<filename>>
Default:
-cache=use
By
default, t_coffee stores in a cache directory, the results of computationnaly
expensive (structural alignment) or network intensive (BLAST search)
operations.
-update
Usage: -update
Default:
turned off
Causes
a wget
access that checks whether the t_coffee version you are using needs updating.
-full_log
Usage: -full_log=<filename>
Default:
turned off
Causes
t_coffee to output a full log file that contains all the input/output files.
-other_pg
Usage: -other_pg=<filename>
Default:
turned off
Some rumours
claim that Tetris is embedded within T-Coffee and could be ran using some
special set of commands. We wish to deny these rumours, although we may admit
that several interesting reformatting programs are now embedded in t_coffee and
can be ran through the –other_pg flag.
seq_reformat: is a versatile reformat
utility. You can get an idea of what it does by typing:
t_coffee –other_pg=seq_reformat
[**]
extract_perl_scripts: will cause t_coffee to
dump all its embedded perl scripts in corresponding files.
t_coffee –other_pg=unpack_all [**]
t_coffee –other_pg=unpack_extract_from_pdb [**]
extract_from_pdb is a useful perl utility we
maintain, meant to parse PDB files. You can also run any of the scripts
embedded in t_coffee by typing:
t_coffee –other_pg=extract_from_pdb
–help [**]
Input
Sequence Input
To
remain compatible with ClustalW, it is possible to indicate the sequences with
this flag
t_coffee
-infile=sample_seq1.fasta [**]
Note: Common multiple sequence alignments format constitute a valid
input format.
Note: T-Coffee automatically removes the gaps before doing the alignment. This behaviour
is different from that of ClustalW where the gaps are kept.
Cf –in from the Method and Library Input section
Usage: -get_type
Default: turned off
Forces
t_coffee to identify the sequences type (PROTEIN, DNA).
Usage:
-type=DNA ¦ PROTEIN¦ DNA_PROTEIN
Default:
-type=<automatically set>
This
flag sets the type of the sequences. If omitted, the type is guessed
automatically. This flag is compatible with ClustalW.
Note: In case of low complexity
or short sequences, it is recommended to set the type manually.