EpDis and MassPred Documentation

1.1 Description

MassPred is a command line (shell) oriented system which enables multiple (massive) and parallel execution of predictors with group FASTA files. Results are filtered and stored in a form suitable for loading into SQL. Also MassPred accepts multiple proteins in the FASTA file. Also, it can be used to execute any command on a set of proteins.

MassPred is a set of tools that provides easy predictors installation, application of predictors to input data and filtering of the results of predictor action. MassPred provides scripts for automated installation of the desired predictors and their preparation for automatic execution. Predictors (any of the previously mentioned three types) or methods for hydropathy calculation are applied to the protein dataset, which can be stored in one or more files or directories. Each file can include one or more proteins in FASTA format. MassPred takes the contents of input files or directories, extracts every single protein and applies the desired predictors to the extracted proteins, creating separate jobs for every pair (protein, predictor). The created jobs can be simultaneously executed on a symmetric multiprocessor computer, or on computers with a multicore/multithread processor architecture. MassPred itself does not perform a parallel execution of a single predictor application to a single protein.

MassPred collects the results and filters them in a CSV file format in order to prepare the results in a form that can be used as an input in a LOAD utility program for loading results in RDBMS tables. By default, the results are filtered for IBM DB2 RDBMS, but can be modified for any other relational database system.

The MassPred mechanism for generating jobs for simultaneous execution is not restricted to supported predictors. In fact, any program that takes a single protein in FASTA format can be massively applied to a set of proteins.

1.1.1 List of supported predictors

1.1.1.1 Disorder predictors

• VSL2b
• IUPred (Long, Short) v1.0
• DisEMBL (Coin coils, Hot loops, Remark465) v1.4
• RONN v3.1
• Predisorder v1.1
• IsUnstruct v2.02
• Disopred v2.43
• OnD-CRF v1.0

1.1.1.2 MHC binding predictors

• NetMHCpan 2.0 v2.0c
• NetMHCpan 2.4 v2.4a
• NetMHCpan 2.8 v2.8a
• NetMHCIIpan 1.0 v1.0b
• NetMHCIIpan 2.0 v2.0b
• NetMHCIIpan 3.0 v3.0c
• NetMHC 3.0 v3.0c
• NetMHC 3.4 v3.4a
• NetMHCII 2.2
• IEDB Tool MHC I (ann, smm, smmpmbec)
• IEDB Tool MHC II (nn_align, smm_align)

1.1.1.3 Hydropathy calculation

Hydropathy of peptides or protein regions is calculated as the sum of hydropathy values of all the AAs, according to chosen scale, divided by the number of residues in peptide (protein region). Methods are based on the following two scales:

• Kyte-Doolittle
• Hopp-Woods

1.1.1.4 Disorder-binding predictors

• ANCHOR v1.0

1.2 Installation

Package installation is done in two steps: MassPred installation and installation of predictors that are necessary for work.

1.2.1 MassPred installation

Unpack MassPred distribution file with command:

tar xzf masspredx.xx.tar.gz.

Name of directory which includes unpacked version should includes only letters, digits, minus, underline, point, or commas. Other symbols in the path may cause a problem in MassPred's work. Recommended directory for installation is /usr/local/masspred.

MassPred directory after unpacking contains sub directory named predictors. This directory includes sub directory source with the following structure:

• clean.sh - shell script which removes all installed predictors;
• data - directory with source of predictors, downloaded from predictors URL;
• data.md5 - MD5 sum of predictors sources files;
• download - directory with shell scripts for automatic download (it is available) of predictor source file;
• download_all.sh - shell script which downloads all available predictors sources files;
• install - directory with shell scripts for installing predictors;
• install_all.sh - shell script which installs all predictors;
• md5sum_data.sh - shell script which calculates MD5 sums of predictors sources files, output must be the same as data.md5;
• patch - directory with patch files for some predictors;
• test - directory with shell scripts for testing predictors;
• test_all.sh - shell script which tests all predictors

Predictors sources files are not included because of the licenses, but some predictors or some of their components are on permanent URL (listed above in list of supported predictors) which is used by download scripts.

1.2.2 Predictors installation

First step in predictor installation process is install the tools necessary for the installation process and operation of predictors. Tools includes specific version of Python and runtime C/C++ libraries. For Ubuntu Linux use the following command:

sudo apt-get install build-essential gawk python-dev tcsh defaultjre libstdc++5 libstdc++5:i386 libstdc++6:i386 libc6:i386 libncurses5:i386 lib32stdc++6 lib32ncurses5 lib32z1 lib32bz21.0

Instruction for the predictors installation are:

1. Open terminal window and go to the predictors directory;
2. Execute source/download_all.sh to download all available predictors;
3. If some of predictors requires additional files, download it as described in its own installation requirements;
4. Optionally execute source/md5sum_data.sh > source/data_new.md5 to calculate MD5 sums;
5. Optionally compare source/data.md5 and source/data_new.md5; generally they should be equal;
6. Execute source/install_all.sh to install all predictors;
7. Optionally execute source/test_all.sh to test all predictor(s) installation;

The specific information about download individual predictors and their test after installation are listed in the rest of this chapter.

1.3 Configuration

Before running MassPred, it is necessary to create a configuration file with lines in the following form:

variable=value

Comments are from # to the end of the line. File configuration.ish in installed directory is an example of configuration file.

Variables that can be put into the configuratio file and their meaning are listed in the table below.

Name	Description
CPU_NUMBER	Specify maximal number of simultaneously executed jobs. Parameter value is number, with default 4.
WORK_SQL	Specify if SQL files with insert statements should be generated. Parameter value is string with default value "no". Possible values are "yes" or "no".
WORK_NUMERIC	Specify generation of alternative form results with numeric value for each AA. Parameter value is string with default value "no". Possible values are "yes"; or "no".
WORK_COMMAND	Specify execution of some command for each protein. Parameter value is string with default value "no". Possible values are "yes"; or "no"
COMMAND	This variable has meaning only if WORK_COMMAND is set. Specify command which executes for each protein. Parameter value is string with default value "-" (when report error, because "-" is nonexistent command).
WORK_HYDRO	Specify generation of table hydro for each AA in protein. Parameter value is string with default value "no". Possible values are "yes"; or "no"
WORK_ANCHOR	Specify execution of ANCHOR predictor. Parameter value is string with default value "no". Possible values are "yes"; or "no"
WORK_DISEMBL	Specify execution of DisEMBL. Parameter value is string with default value "no". Possible values are "yes"; or "no"
WORK_DISOPRED	Specify execution of disopred predictor. Parameter value is string with default value "no". Possible values are "yes"; or "no"
WORK_ISUNSTRUCT	Specify execution of IsUnstruct predictor. Parameter value is string with default value "no". Possible values are "yes"; or "no"
WORK_IUPRED_LONG	Specify execution of iupred predictor, long variant. Parameter value is string with default value "no". Possible values are "yes"; or "no"
WORK_IUPRED_SHORT	Specify execution of iupred predictor, short variant. Parameter value is string with default value "no". Possible values are "yes"; or "no"
WORK_OND	Specify execution of OnD predictor. Parameter value is string with default value "no". Possible values are "yes"; or "no"
WORK_PREDISORDER	Specify execution of predisorder predictor. Parameter value is string with default value "no". Possible values are "yes"; or "no"
WORK_RONN	Specify execution of RONN predictor. Parameter value is string with default value "no". Possible values are "yes"; or "no"
WORK_VSL2	Specify execution of VSL2 predictor. Parameter value is string with default value "no". Possible values are "yes"; or "no"
WORK_NETMHC_1_30C	Specify execution of netMHC predictor, version 3.0c. Parameter value is string with default value "no". Possible values are "yes"; or "no"
NETMHC_1_30C_ALLELE_FILE	This variable has meaning only if WORK_NETMHC_1_30C is set. Specifies file with allele. Path is specified absolute or relative with regard to configuration file. Parameter value is string with default value "NetMhc.3.0c.pseudo".
NETMHC_1_30C_LENGTH_FROM	This variable has meaning only if WORK_NETMHC_1_30C is set. Specifies from which length of peptide predictor works. Parameter value is number, with default 8.
NETMHC_1_30C_LENGTH_TO	This variable has meaning only if WORK_NETMHC_1_30C is set. Specifies to which length of peptide predictor works. Parameter value is number, with default 11.
NETMHC_1_34A_ALLELE_FILE	This variable has meaning only if WORK_NETMHC_1_34A is set. Specifies file with allele. Path is specified absolute or relative with regard to configuration file. Parameter value is string with default value "NetMhc.3.4a.pseudo"
NETMHC_1_34A_LENGTH_FROM	This variable has meaning only if WORK_NETMHC_1_34A is set. Specifies from which length of peptide predictor works. Parameter value is number, with default 8.
NETMHC_1_34A_LENGTH_TO	This variable has meaning only if WORK_NETMHC_1_34A is set. Specifies to which length of peptide predictor works. Parameter value is number, with default 11.
WORK_NETMHC_2_22	Specifies execution of netMHCII predictor, version 2.2. Parameter value is string with default value "no". Possible values are "yes"; or "no"
NETMHC_2_22_ALLELE_FILE	This variable has meaning only if WORK_NETMHC_2_22 is set. Specifies file with allele. Path is specified absolute or relative with regard to configuration file. Parameter value is string with default value "NetMhcII.2.2.pseudo".
NETMHC_2_22_LENGTH_FROM	This variable has meaning only if WORK_NETMHC_2_22 is set. Specifies from which length of peptide predictor works. Parameter value is number, with default 9.
NETMHC_2_22_LENGTH_TO	This variable has meaning only if WORK_NETMHC_2_22 is set. Specifies to which length of peptide predictor works. Parameter value is number, with default 15.
WORK_NETMHC_PAN_1_20C	Specifies execution of netMHCPan predictor, version 2.0c. Parameter value is string with default value "no". Possible values are "yes"; or "no"
NETMHCPAN_1_20C_ALLELE_FILE	This variable has meaning only if WORK_NETMHC_PAN_1_20C is set. Specifies file with allele. Path is specified absolute or relative with regard to configuration file. Parameter value is string with default value "NetMhcPan.2.0c.pseudo".
NETMHCPAN_1_20C_LENGTH_FROM	This variable has meaning only if WORK_NETMHC_PAN_1_20C is set. Specifies from which length of peptide predictor works. Parameter value is number, with default 8.
NETMHCPAN_1_20C_LENGTH_TO	This variable has meaning only if WORK_NETMHC_PAN_1_20C is set. Specifies to which length of peptide predictor works. Parameter value is number, with default 11.
WORK_NETMHCPAN_1_24A	Specifies execution of netMHCPan predictor, version 2.4a. Parameter value is string with default value "no". Possible values are "yes"; or "no"
NETMHCPAN_1_24A_ALLELE_FILE	This variable has meaning only if WORK_NETMHCPAN_1_24A is set. Specifies file with allele. Path is specified absolute or relative with regard to configuration file. Parameter value is string with default value "NetMhcPan.2.4a.pseudo".
NETMHCPAN_1_24A_LENGTH_FROM	This variable has meaning only if WORK_NETMHCPAN_1_24A is set. Specifies from which length of peptide predictor works. Parameter value is number, with default 8.
NETMHCPAN_1_24A_LENGTH_TO	This variable has meaning only if WORK_NETMHCPAN_1_24A is set. Specifies to which length of peptide predictor works. Parameter value is number, with default 11.
WORK_NETMHCPAN_1_28A	Specifies execution of netMHCPan predictor, version 2.8a. Parameter value is string with default value "no". Possible values are "yes"; or "no"
NETMHCPAN_1_28A_ALLELE_FILE	This variable has meaning only if WORK_NETMHCPAN_1_28A is set. Specifies file with allele. Path is specified absolute or relative with regard to configuration file. Parameter value is string with default value "NetMhcPan.2.8a.pseudo".
NETMHCPAN_1_28A_LENGTH_FROM	This variable has meaning only if WORK_NETMHCPAN_1_28A is set. Specifies from which length of peptide predictor works. Parameter value is number, with default 8.
NETMHCPAN_1_28A_LENGTH_TO	This variable has meaning only if WORK_NETMHCPAN_1_28A is set. Specifies to which length of peptide predictor works. Parameter value is number, with default 11.
WORK_NETMHCPAN_2_10B	Specifies execution of netMHCIIPan predictor, version 1.0b. Parameter value is string with default value "no". Possible values are "yes"; or "no"
NETMHCPAN_2_10B_ALLELE_FILE	This variable has meaning only if WORK_NETMHCPAN_2_10B is set. Specifies file with allele. Path is specified absolute or relative with regard to configuration file. Parameter value is string with default value "NetMhcIIPan.1.0b.pseudo".
NETMHCPAN_2_10B_LENGTH_FROM	This variable has meaning only if WORK_NETMHCPAN_2_10B is set. Specifies from which length of peptide predictor works. Parameter value is number, with default 9.
NETMHCPAN_2_10B_LENGTH_TO	This variable has meaning only if WORK_NETMHCPAN_2_10B is set. Specifies to which length of peptide predictor works. Parameter value is number, with default 15.
WORK_NETMHCPAN_2_20B	Specifies execution of netMHCIIPan predictor, version 2.0b. Parameter value is string with default value "no". Possible values are "yes"; or "no"
NETMHCPAN_2_20B_ALLELE_FILE	This variable has meaning only if WORK_NETMHCPAN_2_20B is set. Specifies file with allele. Path is specified absolute or relative with regard to configuration file. Parameter value is string with default value "NetMhcIIPan.2.0b.pseudo".
NETMHCPAN_2_20B_LENGTH_FROM	This variable has meaning only if WORK_NETMHCPAN_2_20B is set. Specifies from which length of peptide predictor works. Parameter value is number, with default 9.
NETMHCPAN_2_20B_LENGTH_TO	This variable has meaning only if WORK_NETMHCPAN_2_20B is set. Specifies to which length of peptide predictor works. Parameter value is number, with default 15.
WORK_NETMHCPAN_2_30C	Specifies execution of netMHCIIPan predictor, version 3.0c. Parameter value is string with default value "no". Possible values are "yes"; or "no"
NETMHCPAN_2_30C_ALLELE_FILE	This variable has meaning only if WORK_NETMHCPAN_2_30C is set. Specifies file with allele. Path is specified absolute or relative with regard to configuration file. Parameter value is string with default value "NetMhcIIPan.3.0c.pseudo".
NETMHCPAN_2_30C_LENGTH_FROM	This variable has meaning only if WORK_NETMHCPAN_2_30C is set. Specifies from which length of peptide predictor works. Parameter value is number, with default 9.
NETMHCPAN_2_30C_LENGTH_TO	This variable has meaning only if WORK_NETMHCPAN_2_30C is set. Specifies to which length of peptide predictor works. Parameter value is number, with default 15.

1.4 Running

MassPred can be started with command in installed directory:
./work.sh <configuration_file> <input>

Input is a directory with FASTA files or a single FASTA file with multiple FASTA formats. For each protein in each FASTA file, MassPred executes required predictors and/or command if the command is given.

1.5 Results

After MassPred's successful ending, the result will be found in the output directory that is printed at the end of execution. The name of the output directory is calculated as the name of the input with added ".out". If COMMAND variable is specified, in the output directory will be found the subdirectory command with the results of the execution command for each protein. Other files are the results of execution requested predictor(s) on input data.

1.5.1 Generated files

The results of the predictor(s) application can be found in the output directory. Depending on which predictors are applied, the appropriate file with the results is generated. The files are compressed with gzip utility in order to save disk space and easier manipulation. The process always generate the LOAD files, while generating the SQL files is optional. The _fail files appear if the error occurred during the (any) predictor execution (the most common example is execution of VSL2b predictor which can not be applied on proteins with letters denotes ambiguous AAs).

Possible files in the output directory:

• epitope_fail.load.gz
  gziped LOAD file with list of netMHC failed predictors. This file is generated if the error occurred in the execution of netMHC predictors.
• epitope_fail.sql.gz
  gziped SQL file with list of netMHC failed predictors. This file is generated if the error occurred in the execution of netMHC predictors.
• region_fail.load.gz
  gziped LOAD file with list of disorder failed predictors. This file is generated if the error occurred in the execution of disorder predictors.
• region_fail.sql.gz
  gziped SQL file with list of disorder failed predictors. This file is generated if the error occurred in the execution of disorder predictors.
• hydro.load.gz
  gziped LOAD file with the results of hydrophobia.
• hydro.sql.gz
  
• epitope_success.load.gz
  gziped LOAD file with the results of netMHC predictors.
• epitope_success.sql.gz
  gziped SQL file with the results of netMHC predictors.
• epitope_success_numeric.load.gz
  gziped LOAD file with numeric results of netMHC predictors. This file is generated if WORK_NUMERIC option is given.
• epitope_success_numeric.sql.gz
  gziped SQL file with numeric results of netMHC predictors. This file is generated if WORK_NUMERIC option is given.
• region_success.load.gz
  gziped LOAD file with the results of disorder predictors.
• region_success.sql.gz
  gziped SQL file with the results of disorder predictors.
• region_success_numeric.load.gz
  gziped LOAD file with numeric results of disorder predictors. This file is generated if WORK_NUMERIC option is given.
• region_success_numeric.sql.gz
  results of disorder predictors. This file is generated if WORK_NUMERIC option is given.

MassPred collects the results and filters them in a CSV file format in order to prepare the results in a form that can be used as an input in a load utility program for loading results in RDBMS tables. By default, the results are filtered for IBM DB2 RDBMS. Structure of the tables that can be used are:

epitope_fail

(

	protein_id		VARCHAR(64),

	protein_reference	VARCHAR(64),

	protein_file_name	VARCHAR(64),

	type 			VARCHAR(32),

	allele			VARCHAR(32),

	length			INTEGER

)



region_fail

(

	protein_id		VARCHAR(64),

	protein_reference	VARCHAR(64),

	protein_file_name	VARCHAR(64),

	type			VARCHAR(32)

)



hydro

(

	protein_id		VARCHAR(64),

	protein_reference	VARCHAR(64),

	protein_file_name	VARCHAR(64),

	position		INTEGER,

	aa			CHAR(1),

	hydro_kd		DECIMAL,

	hydro_hw		DECIMAL

)



epitope

(

	protein_id		VARCHAR(64),

	protein_reference	VARCHAR(64),

	protein_file_name	VARCHAR(64),

	position		INTEGER,

	epitope			VARCHAR(32),

	pos			INTEGER,

	core			VARCHAR(32),

	aff_log			DECIMAL,

	aff			DECIMAL,

	rank			DECIMAL,

	binding			CHAR(2),

	type			VARCHAR(32),

	allele			VARCHAR(32),

	length			INTEGER

)



epitope_numeric

(

	protein_id		VARCHAR(64),

	protein_reference	VARCHAR(64),

	protein_file_name	VARCHAR(64),

	position		INTEGER,

	epitope			VARCHAR(32),

	pos			INTEGER,

	core			VARCHAR(32),

	aff_log			DECIMAL,

	aff			DECIMAL,

	rank			DECIMAL,

	binding			CHAR(2),

	type			VARCHAR(32),

	allele			VARCHAR(32),

	length			INTEGER

)



region

(

	protein_id		VARCHAR(64),

	protein_reference	VARCHAR(64),

	protein_file_name	VARCHAR(64),

	begin			INTEGER,

	end			INTEGER,

	order			CHAR(1),

	type			VARCHAR(32)

)



region_numeric

(

	protein_id		VARCHAR(64),

	protein_reference	VARCHAR(64),

	protein_file_name	VARCHAR(64),

	position		INTEGER,

	aa			CHAR(1),

	value			DECIMAL,

	order			CHAR(1),

	type			VARCHAR(32)

)

							

1.5.2 Command

If the option WORK_COMMAND is set, then in subdirectory command, for each protein from input FASTA files, three files with file name in form: <name>.<number>.<suffix>, "name" is the name of input file and "number" is the position of the protein in FASTA file. Meanning of the value for suffix are:

• "rc" - file with result code of executed command,
• "out" - file with standard output of executed command,
• "err" - file with standard error output of executed command.

The command gets from standard input content of FASTA file (which refers to the specific protein in the order).
Before the execution of command, the next environment variables are set:

• MASSPRED_INPUT - name of input FASTA file,
• MASSPRED_INPUT_POSITION - position of protein in input FASTA file,
• MASSPRED_OUTPUT - prefix of file name output files,
• MASSPRED_FASTA_DB - first field from header of input FASTA file,
• MASSPRED_FASTA_ID - second field from header of input FASTA file,
• MASSPRED_FASTA_REFERENCE_DB - third field from header of input FASTA file,
• MASSPRED_FASTA_REFERENCE - fourth field from header of input FASTA file,
• MASSPRED_FILE_NAME - base part of input FASTA file name.

1.6 Tests

Calling MassPred is very simple. On the command line from directory where MassPred is unpacked (for example in directory /usr/local/masspred), just execute work.sh script:

./work.sh my_directory/configuration.ish my_directory/my_file.faa

Where configuration.ish is configuration file described in chapter 1.3, and my_file.faa is file that contains FASTA format of protein(s) or the directory with files containing FASTA protein code in FASTA format. For example, the following commands can be used for the test of successful installation of MassPred:

./work.sh test/configuration.ish test/iedb_90149017.faa
./work.sh test/configuration.ish test/iedb_9000.faa
./work.sh test/configuration.ish test/dir
./work.sh test/configuration.ish test/cancer.faa
./work.sh test/configuration.ish test/p03211.faa
./work.sh test/configuration.ish test/NC_000913.faa

1.7 Uninstallation

For uninstalling MassPred just remove the directory in which MassPred is installed.

1.8 Update

The longer and more secure way is to uninstall MassPred first and then install it again.

The shorter way allows retain installed predictors. Before removing the directory containing the MassPred, move the directory predictors (with installed predictors) to a temporary location. Remove the Masspred installation directory, and after that unpack again MassPred (in the same directory) and replace directory predictors with the previous version (from the temporary location).

It is important that the absolute path to the predictors directory is not changed. In order to ensure that, it is essential to always install the MassPred in the same directory (for example /usr/local/masspred).

1.9 Possible problems

When the execution is normal, the MassPred cleans all the temporary directories, but in the case of execution terminates abnormally, it is possible that some temporary directories will not be properly removed. In this case, it is necessary to manually remove all temporary directories, which have form:

/tmp/masspreddir <number> <number>

When the execution is normal, the MassPred cleans all the temporary directories, but in the case of execution terminates abnormally, it is possible that some temporary directories will not be properly removed.

1.10 Application example

MassPred, is a script oriented system. It is started by running the script work.sh with two input parameters: name of the MassPred application configuration file and the name of the fasta file or directory which includes fasta files. For example,

/usr/local/masspred/work.sh configuration.ish p03211.faa

The configuration file (configuration.ish) includes parameters related to the current computer system (for example, the number of concurrent CPUs), and parameters related to supported predictors. An example of a configuration file is:

CPU_NUMBER=8
WORK_HYDRO=yes
WORK_ISUNSTRUCT=yes
WORK_VSL2=yes
WORK_NETMHC_1_34A=yes
NETMHC_1_34A_ALLELE_FILE=NetMhc.3.4a.pseudo
NETMHC_1_34A_LENGTH_FROM=8
NETMHC_1_34A_LENGTH_TO=9
...

File p03211.faa includes the fasta version of protein(s) on which the predictors are to be applied. The file can include one or more proteins in fasta format. For example,

>gi|119110|sp|P03211.1|EBNA1_EBVB9 RecName: Full=Epstein-Barr nuclear 

antigen 1; Short=EBNA-1; Short=EBV nuclear antigen 1

MSDEGPGTGPGNGLGEKGDTSGPEGSGGSGPQRRGGDNHGRGRGRGRGRGGGRPGAPGGSGSGPRHRDGV

RRPQKRPSCIGCKGTHGGTGAGAGAGGAGAGGAGAGGGAGAGGGAGGAGGAGGAGAGGGAGAGGGAGGAG

GAGAGGGAGAGGGAGGAGAGGGAGGAGGAGAGGGAGAGGGAGGAGAGGGAGGAGGAGAGGGAGAGGAGGA

GGAGAGGAGAGGGAGGAGGAGAGGAGAGGAGAGGAGAGGAGGAGAGGAGGAGAGGAGGAGAGGGAGGAGA

GGGAGGAGAGGAGGAGAGGAGGAGAGGAGGAGAGGGAGAGGAGAGGGGRGRGGSGGRGRGGSGGRGRGGS

GGRRGRGRERARGGSRERARGRGRGRGEKRPRSPSSQSSSSGSPPRRPPPGRRPFFHPVGEADYFEYHQE

GGPDGEPDVPPGAIEQGPADDPGEGPSTGPRGQGDGGRRKKGGWFGKHRGQGGSNPKFENIAEGLRALLA

RSHVERTTDEGTWVAGVFVYGGSKTSLYNLRRGTALAIPQCRLTPLSRLPFGMAPGPGPQPGPLRESIVC

YFMVFLQTHIFAEVLKDAIKDLVMTKPAPTCNIRVTVCSFDDGVDLPPWFPPMVEGAAAEGDDGDDGDEG

GDGDEGEEGQE

>gi|89902357|ref|YP_524828.1| 50S ribosomal protein L7/L12 [Rhodoferax ferrireducens T118]

MAFDKDAFLTALDSMTVMELNDLVKAIEEKFGVSAAAMSAPAAGGAVAAVAEEKTEFNVVLLEAGAAKVS

VIKAVREITGLGLKEAKDMVDGAPKNVKEGVSKVDAEAALKKLLDAGAKAELK

.............

							

After execution, Masspred creates a directory with the added suffix .out to the name of the input file. The directory contains gzipped files with results for every required prediction. For example, for the part of the configuration file shown above, a directory p03211.faa.out was created with the following files:

• epitope_success.load.gz - gzipped file with the predicted data from the epitope predictors
• hydro.load.gz - gzipped file with calculated hydrophobicity values for every AA
• region_success.load.gz - gzipped file with the predicted data from the disorder predictors

All files are in TAB separated format and prepared for loading into RDBMS. For example, the content of region_success.load file is:

119110		P03211.1	p03211	1	413	D	IsUnstruct

119110		P03211.1	p03211	414	414	O	IsUnstruct

119110		P03211.1	p03211	415	480	D	IsUnstruct

119110		P03211.1	p03211	481	541	O	IsUnstruct

119110		P03211.1	p03211	542	554	D	IsUnstruct

119110		P03211.1	p03211	555	613	O	IsUnstruct

119110		P03211.1	p03211	614	641	D	IsUnstruct

89902357	YP_524828.1	p03211	1	55	D	IsUnstruct

89902357	YP_524828.1	p03211	56	84	O	IsUnstruct

89902357	YP_524828.1	p03211	85	123	D	IsUnstruct

119110		P03211.1	p03211	1	479	D	VSL2b

119110		P03211.1	p03211	480	497	O	VSL2b

119110		P03211.1	p03211	498	498	D	VSL2b

119110		P03211.1	p03211	499	541	O	VSL2b

119110		P03211.1	p03211	542	553	D	VSL2b

119110		P03211.1	p03211	554	606	O	VSL2b

119110		P03211.1	p03211	607	641	D	VSL2b

89902357	YP_524828.1	p03211	1	5	D	VSL2b

89902357	YP_524828.1	p03211	6	36	O	VSL2b

89902357	YP_524828.1	p03211	37	44	D	VSL2b

89902357	YP_524828.1	p03211	45	84	O	VSL2b

89902357	YP_524828.1	p03211	85	123	D	VSL2b

							

2.1 Requirements

In order to install and use EpDis application Oracle Java should be installed in the system, at least JRE 6. Java Development Kit (JDK) can be installed in the following way:

1. download latest JDK distribution from http://www.oracle.com/technetwork/java/javase/downloads/index.html for your Linux distribution, as tar.gz archive and unpack it in some dedicated folder,
2. go to folder /usr/bin and create symbolic link with following command:

   sudo ln -s /home/Java/jdk1.6.0_29/bin/java java

   where "/home/Java/jdk1.6.0_29" is path to the folder where JDK distribution was unpacked.

2.2 Installation

To install application execute following command:

java –jar epdis-install.jar

and follow these steps:

Remark: For further notice let %EPDIS_HOME% be the path to installation folder, as it was specified in Step 2. (in this example /home/daca/EPDIS).

2.3 Uninstallation

To uninstall application, go to folder %EPDIS_HOME%/Uninstaller/ and execute following command:

java –jar uninstaller.jar

2.4 Configuration

After instalation, there are customizable configuration files in %EPDIS_HOME%/config/ folder , which can be changed and accommodated for local environment:

epdis.properties

Contains global application configuration with following properties:

Name	Description
DataManagementModule.enabled	Whether data management module is enabled or not. By default it is disabled. Enabling this module it is possible to load proteins from database and execute prediction methods on it, as well as saving of prediction results directly into the database through MassPred component. Possible values are true/false.
ExecutionModule.show_not_configured_methods_on_startup	Whether notification about not properly configured methods should be shown on application startup or not. (see next paragraph). Default value is true. Possible values are true/false
MassPredModule.enabled	Whether the MassPred is enabled or not.

methods.properties

Contains properties of each prediction method. EpDis application comes with support for following methods:

Epitope predictors
NetMHCpan versions: 2.0, 2.4, 2.8	http://www.cbs.dtu.dk/services/NetMHCpan/
NetMHCIIPan versions: 1.0, 2.0, 3.0	http://www.cbs.dtu.dk/services/NetMHCIIpan/
NetMHC3.0, 3.4	http://www.cbs.dtu.dk/services/NetMHC-3.0/
NetMHCII2.2	http://www.cbs.dtu.dk/services/NetMHCII/
IEDB_Tool_MHC_I (ann, smm, smmpmbec)	http://tools.immuneepitope.org/mhci/download/
IEDB_Tool_MHC_II (nn_align, smm_align)	http://tools.immuneepitope.org/mhcii/download/
Disorder predictors
VSL2b	http://www.ist.temple.edu/disprot/predictorVSL2.php
Predisorder	http://casp.rnet.missouri.edu/predisorder.html
IUPred	http://iupred.enzim.hu/
RONN	http://www.bioinformatics.nl/~berndb/ronn.html
IsUnstruct	http://bioinfo.protres.ru/IsUnstruct/
DisEMBL	http://dis.embl.de/
OnD-CRF	http://babel.ucmp.umu.se/ond-crf/
Disopred	http://bioinf.cs.ucl.ac.uk/psipred/?disopred=1
Disorder-binding predictor
ANCHOR	http://anchor.enzim.hu/
Hydrophobicity scales
Kyte Doolitle	http://www.vivo.colostate.edu/molkit/hydropathy/scales.html
Hopp-Woods	http://www.vivo.colostate.edu/molkit/hydropathy/scales.html

For each method, the set of its properties is defined in the following way:

Name	Value	Description
netMhcPanPredictor.name	NetMhcPan	The name of the prediction method. Must be unique.
netMhcPanPredictor.programHome	/home/daca/dev/netMHCpan-2.4	Path to the method installation on local file system. *Normally this is the only value that should be changed.
netMhcPanPredictor.programCommand	${netMhcPanPredictor.programHome}/netMHCpan	Program execution command. Usually should not be changed. Value of the expression ${netMhcPanPredictor.programHome} represents value of netMhcPanPredictor.programHome property.
netMhcPanPredictor.allelesSource	${netMhcPanPredictor.programHome}/data/MHC_pseudo.dat	Path to the file with Allele&s definition. Usually should not be changed

Usually, only value of netMhcPanPredictor.programHome should be changed to fit local environment. During startup, application will check this configuration file, and load only those methods which are configured properly, i.e. it will load method if the path to the corresponding program installation is valid. For those methods which are not configured properly, application will notify user if value of the property ExecutionModule.show_not_configured_methods_on_startup is set to true.

jdbc.properties

Contains properties related to database connectivity:

Name	Description
db.driverClass	JDBC driver class, should not be changed
db.port	Database port
db.host	Database host
db.name	Database name, containing table definitions for application purposes
db.default_schema	Database schema
db.url	JDBC url, composed of values of other properties, should not be changed
db.username	User name
db.password	Password

masspred.properties

Name	Description
massPred.Home	Path to the method installation on local file system
massPred.Command=./work.sh	Program execution command. Should not be changed.

2.5 Running

Running of EpDis can be performed in two following ways:

• from command line, executing java -jar epdis-4.0.0.jar command from installation folder, or
• by clicking on the shortcut icon from the start menu, created during installation (Step 4.)

2.5.1 Making predictions

After successful startup the following screen will appear:

User interface is divided into tabs (one or two based on configuration). 'Execution' tab is composed of four parts:

1. Protein selection (part A):
   1. from database
   2. from file in FASTA format
   3. from UNIPROT database, by specifying protein identifier
   4. as raw protein sequence
2. Disorder and disorder-binding predictor selection (part B) - it is possible to choose one or more predictors to make predictions.
3. Epitope predictor selection (part C) - single selection of different prediction method. In this example the NetMHCIIPan3.0 predictor is selected, and specified parameters: allele DRB1_0101, peptide of length 9 and default treshold. For MHC binding predictors results are values that represent the binding affinity of the peptides. The choice of an appropriate threshold is a critical task and should be chosen carefully. Inclusion of an experimental determined data about T cell epitopes and MHC binders might help in choosing an appropriate threshold value (see bellow).
4. Hydropathy calculation (part D) - selection of hydropathy scale and peptide length.

At least one method in either part B, C or D should be specified, or an arbitrary combination of predictors from these parts.

It is possible to include in the display, beside predictors results, experimental data about secondary structures and T-cell epitopes. Entering experimental data is based on the addition of an appropriate xml file whose structure is in accordance with the pre-defined XML scheme:

• epdis-experimental-disorder.xsd
• epdis-experimental-epitope.xsd

Adding, deleting and overviewing of already added experimental data can be done by choosing corresponding action from 'Experimental' menu item. EpDis displays added experimental data in a form of table:

Predictions for selected set of predictors are done by clicking on the button 'Execute'. The panel with visualized results is presented:

All the colors and annotations can be adjusted either in the configuration files or during runtime. It is possible to choose what is to be presented, whether intervals of ordered or of disordered parts of a protein. Each displayed disorder predictor result may be excluded/included from the graphic display (by checking/unchecking the check box next to the name of predictor in the top right table). The native result in the text form of each predictor can be displayed by clicking on predictor name. Each graphic can be individually exported to various graphic formats, such as PNG, PDF, SVG, and EPS. The experimentally validated T cell epitopes and MHC binders can be displayed in all three graphics or only 'epitope part D'. All the experimental data is displayed in in the actual length of the epitope and in the actual position within protein sequence.

2.5.2 MassPred integration

MassPred is fully integrated into EpDis tool through 'MassPred module'. This module enables the running of the MassPred tool from graphic mode. Turning on this component is optional and configurable. Apart from extending the functionalities of EpDis, the main purpose of this module is to facilitate the use of the MassPred tool itself, by allowing the user to specify its parameters simply by selecting the target predictors. The MassPred module has a predefined default configuration, but it allows storing and loading of a custom configuration as well. After execution of MassPred, the output is redirected and displayed to the user in the separate window. The interface for the MassPred module is shown in the following picture:

2.6 Database tables design

Apart from before mentioned tables needed for storing data from output of MassPred, following DDL should be executed to create all EpDis-related tables:

CREATE TABLE EPDIS_CACHE (

		KEY			VARCHAR(70) NOT NULL, 

		PREDICTION_RESULT	BLOB(1048576) INLINE LENGTH 164 NOT NULL,

		PRIMARY KEY (KEY)

);



CREATE TABLE PROTEIN (

		ID			VARCHAR(20) NOT NULL, 

		SEQUENCE		VARCHAR(10000) NOT NULL, 

		LENGTH			SMALLINT NOT NULL, 

		FASTA_HEADER		VARCHAR(500) NOT NULL, 

		ORD_NUM_IN_DISPROT	VARCHAR(20) NOT NULL,

		PRIMARY KEY (ID)

);

							

ER diagram describing MassPred/EpDis entities is given in the following image:

If 'DatabaseModule' is enabled tables 'EPITOPE_NUMERIC' and 'REGION_NUMERIC' are used as first level cache during predictors execution. Table 'EPDIS_CACHE' is used as the second level cache.