Back to Marcoil's main page



This version might be outdated, please check with the README included in the downlodable program's distribution. Here first the instructions for downloading and uncompressing.



is a hidden MARkov model for predicting existence and localisation of coiled-coil domains. Questions and comments are welcome, email me at Please report bugs.



After downloading a copy of the file Marcoilcode.tar.gz in an appropriate directory,

extract with the commands

gunzip Marcoilcode.tar.gz

tar -xvf Marcoilcode.tar

This command creates a directory MARCOIL containing all the files;

cd into the new directory and compile the code by typing make. The makefile calls the gcc compiler and the needed libraries. Various files should appear, including the program MARCOIL and the most recent README document.


Back to Marcoil's main page


Nonetheless, in the following a description on how to use Marcoil.



From the MARCOIL directory, run the program on the file SEQUENCES/seqfile2 by typing:

marcoil +cdlsS SEQUENCES/seqfile

and check out the appearance of four result files in the subdirectory Outputs and their content.

The 5 options +cdlsS in the above command do the following:

c the output file CompactProfile is produced

d the output file Domains is produced

l the output file ProbList is produced

s the output file ProbPerState is produced

S the protein Sequence is also written to result files

Marcoil always rewrites the same output files, so to keep results they have to be renamed or moved. When the program has finished, a message is written to stdout, as are some type of problems or errors, if encountered.



Marcoil takes its parameters from a bunch of files all located in the same subdirectory Inputs. This should not be modified. If you run the PSSM algorithm by using the -C option, then the option +s is inactive and the output files are called CompactProfilePSSM, DomainsPSSM and ProbListPSSM.

Running the program under UNIX or LINUX, follow the usual rules for adresses. For example, start the program from its directory with:

./ marcoil

the program looks for SEQUENCES/seqFile and tries to use this file.

The output should be self-explanatory.



Please use only fasta or multifasta format, as only these were tested. The format is not checked by the program and you could get false results.

The number of sequences is unlimited.

Only the 20 standard amino acids are represented in the model, other letters are interpreted as unknown amino acids.

Lowercase is converted to uppercase, numbers, spaces, newlines, tabs and other special signs in the sequence are ignored.

The program should be able to handle an input with the folling schematic structure as examle:

> name of first sequence (name is optional)
FIRST SEQUENCE perhaps with digits
61 continuation of sequence
>name of following sequence
SEQUENCE perhaps with s p a c e s and newlines

61 continuation of sequence




marcoil <-evwHLCI> <+cdlsS> <-E filename> <-T filename> <-c number> <+c number>

The order of the <...> blocks is irrelevant.


a) algorithmic / file / parameter options
[ -H ]
Read transition parameters from the file R3.transProbHigh This is the default. We call it the MARCOIL-H predictor
[ -L ]
Read transition parameters from the file R3.transProbLow This is what we call the MARCOIL-L predictor.
[ -E emfileName]
Read emission parameters from the user's file emfileName
[ -T trfileName]
Read transition parameters from the user's file trfileName
[ -m trfileName]
Base emission parameters on the file R5.MTK or R5.MTIDK (if -i), except state 0, based on R2.emissProb (or the -E file)
[ -i ]
Use the MTIDK matrix instead of MTK (active only under -P or -m) As file name the adress relative to the current directory has to be used.
[ -P ]
Run the PSSM28 algorithm instead of MARCOIL, with the MTK matrix (default, file R5.MTK) or the matrix specified with the -E option The first is equivalent to COILS28 / MTK (up to rounding errors)

b) numeric
[ -c probabilityCutoff]
Results are written only for positions with coiled-coil probability equal or superior to probabilityCutoff
[ -t probabilityThreshold]
For the parsing, for defininf domain borders, use the threshold probabilityThreshold instead of the default values kParsingthreshold, defined in globals.h

c) output options
[ +d ] [ +l ] [ +m ] [ +s ] [ +S ]
Explained above.
[ -v ]
Run in verbose mode. A fair bit of output will be printed to the screen to follow execution. Useful for tracing down bugs.


Back to Marcoil's main page


This page was last modified Thursday, 24-Jan-2002 17:45:48 EST