Home

File

1. e PAUDA User provided reference data in faa format is required for taxonomical and functional classification see CREATING VIRAL DATABASE The PAUDA index is generated by buildIndexes py PAUDA relies on the availability of the Bowtie2 software see INSTALLATION By default VirusLand executes PAUDA using the slow parameter while slower albeit significantly faster than BLAST based searches still less than 1 hour for 1 million reads than the alternative fast more reads are assigned This can be modified by changing the call in the pauda py script Output is produced within the pauda folder For further details regarding PAUDA please refer to the software s publication http www ncbi nim nih gov pubmed 23658416 e Lambda A second more expedient means of classification is provided with the addition of the Lambda classification package Lambda is a local sequence aligner similar to PAUDA in that it uses read mapping approaches to classify reads in a faster manner that BLAST while not sacrificing sensitivity Per Lambda benchmark tests http www ncbi nim nih gov pubmed 25161219 Lambda outperforms produces results more akin to BLAST based searches yet at a fraction of the time PAUDA The index required for running Lambda is generated by buildindexes py Output is produced within the lambda folder VirusLand v1 0 User s Manual Analysis and Visualization Three analysis tools are included in the VirusLand Tax
2. indexes Index files generated by VirusLand prior to runs faa reference faa index and Lambda index gbk reference gbk index paudaindex PAUDA index Note entire folder must be deleted before recomputing index taxindex taxonomy index seqs Optional location to put input sequences in directory location of sequences is still required to be specified as parameters in command line VirusLand v1 0 User s Manual CREATING VIRAL DATABASE A local copy of the reference database must be created e faa and gbk files are required for all sequences species genomes included in the database Such files can be downloaded from NCBI s ftp site ftp ftp ncbi nim nih gov genomes Viruses by selecting the current all faa tar gz and all gbk tar gz files e Folders must be uncompressed Subfolder structure can be maintained although not mandatory e The config file will request the root folder for the faa and gbk files PRE PIPELINE DATA PREPARATION e preCompiler py The C and Java modules provided in the virusLand package must be compiled before the pipeline can be run A Python script has been provided to perform the compilations called preCompiler py This module goes through the files listed in the uncompiled folder and creates executables for each in the src folder Command python preCompiler py e buildindexes py Some of the programs used in the VirusLand package require reference indexes datasets to be computed In order
3. species for which one or more hits were found is able to be visualized Figure 2 shows an example of one such visualization Charts may be copied to the clipboard or saved in png format by right clicking on the graph Furthermore the graph is interactive users can click on a bar to identify the number of hits and gene s NCBI accession number Figure 2 shows an example HitViz visualization A sample file is included within the src javaVisuals folder Enterobacteria phage T4 NC_000866 of Hits amp Ln ah 5 ih buo i it a o 20 An 60 100 120 14 16 18 200 2230 4i 6i gt A Genes 8 umber of Hits Figure 2 Example of visualization generated by HitViz for hits to a single genome Statistical Data Two tab delimited text files are also generated by VirusLand and contained within the stats folder of the run The genome coverage statistics file file name indicator of coverage includes the number of proteins per genome number of hits total to the genome and of the proteins having hits The hits by protein file file indicator of hits by protein includes the number of hits for each gene for each genome included in the reference database genomes having no hits are not listed VirusLand v1 0 User s Manual RUNNING VIRUSLAND PY AS A BATCH PROCESS It is possible to run multiple analyses back to back without interruption via a batch file This is advantageous for users attempting to analyze multipl
4. to reduce the time and memory required for each run an indexing script called buildIndexes py has been created which is run independently of the VirusLand pipeline This script is required to be run before the analysis can be performed but the indexes only need to be recomputed if the reference dataset faa and gbk files has changed between analyses In brief the script creates a concatenated list of GBK files as well as two files listing the local directories of the FAA and GBK files These files are then used to create the PAUDA and Lambda indexes from the reference dataset This reference dataset must be obtained by the user and placed at the directory referenced by the user in the config vl file Lastly another index file a temporary file is created to be used for statistical analysis later in the pipeline Command python buildIndexes py config vl VIRUSLAND MODULES Assembly Velvet The first step of the VirusLand pipeline is the assembly of the FASTQ files using the Velvet assembler to produce contigs This module incorporates the existing Velvet assembler tool Assembly velvet py makes two calls one to velveth and the second to velvetg The basic parameters for Velvet are programmable by the user through filling out the config vl file provided in the src folder These parameters are specific to the velveth process The possible user defined parameters are VirusLand v1 0 User s Manual e format Supported format types inc
5. VIRUSLAND DOCUMENTATION TABLE OF CONTENTS Installation 2 Introduction 3 Objectives 3 Overview of VirusLand 3 Terminology 4 Configuration Module and config vl file 4 Directory Structure 5 Creating Viral Database 6 Pre Pipeline Data Preparation 6 VirusLand Modules 6 Assembly Velvet 6 Gene Prediction EMBOSS getorf 7 Classification 7 PAUDA Lambda Analysis and Visualization 8 Batch file processing 9 Documentation as of 06 01 2015 VirusLand v1 0 User s Manual INSTALLATION System Requirements e nix OS e Minimum RAM requirement 16GB dependent upon Velvet needs function of k number of reads and read size e Java runtime environment version 1 6 or higher e Tested in BioLinux7 and Ubuntu 12 04 1 LTS 64 bit OS Required Compilers Interpreters and Software Compilers typically distributed in nix OS javaC g and python Software Velvet https github com dzerbino velvet tree master e Reference Velvet s README txt for instructions for compiling e VirusLand assumes location of Velvet is within the bin directory e Tested using version 1 0 19 current as of June 1 2015 Emboss http emboss sourceforge net download e Reference Emboss site listed here for instructions regarding unpacking and compiling e VirusLand assumes location of Emboss is within the bin directory e Tested using version 6 6 0 current as of June 1 2015 Lambda http www segan de projects lambda e Reference Lambda site for instruc
6. a Visualization Krona HitViz and Statistical Data Krona Krona http sourceforge net p krona home krona is hierarchical data visualization tool allowing the user to zoom through pie charts Here Krona is used to visualize relative abundances of metagenomic classifications VirusLand creates the Krona formatted file which can be visualized using Microsoft Excel Template for Windows and Mac or via KronaTools for Mac and Linux via the ktlmportText script Please refer to the Krona site for instructions regarding obtaining as well as generating the interactive chart Note Krona is not included in the installation of VirusLand The input file for Krona either template or KronaTools is contained within the stats folder of the run and includes the file name indicator of KronaStats HitViz This tools was developed as part of VirusLand allowing the user to visualize the coverage of hits to individual genomes This tool allows the user to create graphs of the presence absence of a particular species or near neighbor species This Java module is automatically executed as part of the complete pipeline it can however also be run post processing src javaVisuals virusGraphics jar The input for HitViz is a text file containing the species proteins in the genome and number of hits per annotated coding region This file created by VirusLand is contained within the stats folder of the run and includes the file name indicator of HV All
7. e datasets at once The process is simple and involves creating a text file in which each line contains a single command line call of virusLand py with the necessary parameters Example batch file python virusLand py runl seql seq2 configFileLocation python virusLand py run2 seq3 segq4 configFileLocation python virusLand py run3 seq5 seq6 configFileLocation The run names must be unique for each run any runs with the same name will be overwritten as VirusLand uses the runName parameter to generate the run folder The user can create multiple config vl files with different names and parameters which can then be used for separate runs Once this text file has been created and saved you will have to modify the permissions so it is executable In the terminal use the command chmod 755 lt batchFileLocation gt to make the file executable The file can then be run in the terminal by simply typing the location of the file or fileName if you are in the directory containing the batch file
8. forward to use for those with little programming or scripting knowledge while still being open and extend able to those with more computational knowledge This document starts with the objectives of Virus Land and an overview of modules Source code for the modules is available in the src uncompiled folder within the installation package OBJECTIVES e Developing a tool for automated analysis of viral sequences from raw sequence reads to taxonomic and functional classification e Constructing a flexible software that can be customized with the user preferences e Reusing existing software as well as implementing new modules to generate a complete viral analysis tool e Implementing a software that can be run locally in a reasonable amount of time OVERVIEW OF VIRUSLAND There are four major steps in the pipeline as indicated by the dashed boxes in Figure 1 Assembly Gene prediction Classification and Analysis and visualization Assembly l Velvet TtTrttrtrtrrirtrcrtitrtt t Ttr rtitrrtrCrtrtticwurtrrcrrietrtricetreuvertettrt pete f oT ttrtrrtritrrtiticrtrtrttgrTtTrtTrTrttrtItCcrcr rcQ CC CCR ETIT TEIETETE j RawReacs Gene Prediction Raw Reads 3 getorf il m en momoe oe o o e es e E e E E i Reference Mapper 7 Library Prep PAUDA Taxa Visualization Statistical Krona HitViz Data Figure 1 Pipeline for analysis of viral metagenomic sequences using VirusLa
9. he MAXKMERLENGTH set while compiling Velvet VirusLand v1 0 User s Manual The default MAXKMERLENGTH is 31 see Velvet documentation for how to change this setting Implementing tools such as VelvetOptimizer prior to executing VirusLand can assist in identifying if this value should be increased Example Config file VELVET FORMAT fasta VELVET KMER 31 VELVET READTYPE shortPaired FAA DIR home my path virusLand seqs GBK DIR home my path virusLand seqs PAUDA LOCATION home my path pauda 1 0 1 pauda bin STOP DIRECTORY STRUCTURE virusland src This folder contains all of the pipeline modules python C and Java Note the Velvet EMBOSS PAUDA and Lambda packages should be installed in the bin directory on the user s computer as noted in the Installation uncompiled This folder contains the C and Java source code runs This folder includes the results of analysis runName Each run of VirusLand is written to a folder indicated by the user via a command line parameter when invoking virusLand py assembled Velvet assembled contigs kronaFiles Statistics file for Krona visualizer via MS Excel or KronaTools lambda output files from Lambda ORFs EMBOSS predicted ORFs pauda output files from PAUDA stats statistics generated from taxonomic and functional classification these output files are required for visualization in addition to those generated to assist in user analyses
10. lude fasta fastg fasta gz fastq gz sam bam eland and gerald e kmer corresponds to the length in base pairs of the words being hashed refer to the Velvet Manual for further details e read type Supported types include shortPaired short2 same as short but for separate insert size libraries shortPaired2 long for Sanger 454 etc longPaired The default parameters are used for the call to velvetg exp_cov auto Additional parameters applicable to both velveth and velvetg can be specified by the user by adapting the velvet py script For in depth details regarding the Velvet software and parameters the user is referred to the online manual at https www ebi ac uk zerbino velvet Velvet output and intermediate files are written to the assembled folder All the files except for the file containing the contigs are deleted after assembly in an attempt to save memory they often represent a large portion of the disk space requirements for Velvet Gene Prediction EMBOSS getorf After the contigs are generated in the assembly step they are examined to predict coding regions both in the forward and reverse directions 6 frames The current implementation of this module employs the getorf function from the EMBOSS package ORFs are written to the ORFs folder Classification There are presently two options for classification Additional modules can easily be plugged in by adapting the python modules and parser code cFileParse3 cpp
11. nd Blue boxes indicate incorporation of existing tools Black shapes indicate results software developed in this initiative Tan shapes indicate user supplied data VirusLand v1 0 User s Manual A Python module was developed such that via the use of a single config file a user can explicitly define parameter values for the analysis With complete configuration file this module is able to execute the entire workflow with minimal interruptions The program runs from beginning to end upon invocation The program also saves all the intermediate files generated in each step of the process such that the user can access intermediary files e g contigs ORFs etc for independent analysis Analysis of large datasets can be executed in hours on a 64 bit UNIX operating system with at least 16GB of RAM Minimum memory requirements are dependent upon Velvet and a function of k number of reads and read size TERMINOLOGY e Contig A contig refers to overlapping sequence data reads e k mer The term k mer typically refers to all the possible substrings of length k that are contained in a string The amount of possible k mers within a sequence of a given length L is 2L k 1 assuming the sequence is double stranded whilst the amount of possible k mers given n possibilities 4 in the case of DNA is n Velvet utilizes the parameter k in construction of a de Bruijn graph to assemble short reads CONFIGURATION MODULE AND CONFIG VL FILE The c
12. onfiguration module acts as a parser which loads the input sequences and various parameters as provided by the user in the config vl file It is run at the beginning of the modules preCompiler py and buildIndexes py as well as at the beginning of the VirusLand pipeline It is invoked by the aforementioned modules and VirusLand script virusLand py the user does not need to run the module Essential to the configuration and subsequent execution of VirusLand is the config vl file This file is required a sample text file is included within the src folder in the installation package The following illustrates the format of the config file square brackets indicate the required information and should remain in the config file Config file format VELVEI FORMAT lt fa fasta raw fasSta gz fastq gz raw gz sam bam fmtAuto gt VELVET KMER lt kmer size gt VELVET READTYPE lt shortPaired shortPaired2 short short2 long longPaired reference gt FAA DIR lt directory folder containing reference proteins as faa files gt GBK DIR lt directory folder containing reference viral genomes as gbk files gt PAUDA LOCATION lt directory location of PAUDA software gt STOP Each line of the config vl file should have an entry after the header bracketed label if not the analysis will not be completed The first three headers are specific to the parameters required for Velvet assembly k mer size is limited to t
13. tions regarding unpacking and compiling e VirusLand assumes location of Lambda is within the bin directory e Tested using version 0 4 7 current as of June 1 2015 Bowtie2 http bowtie bio sourceforge net bowtie2 index shtml e Reference Bowtie2 site for instructions regarding unpacking and compiling e VirusLand assumes location of Bowtie2 is within the bin directory e Adding Bowtie 2 directory to your PATH environment variable e Tested using version 2 1 0 current as of June 1 2015 PAUDA http ab inf uni tuebingen de software pauda e Application must be unzipped VirusLand assumes location is within the bin directory e PAUDA is dependent upon the Bowtie2 software Bowtie2 must be installed as well e Tested using version 1 0 1 current as of June 1 2015 Each tool should be added to your PATH environment variable Note the Bio Linux OS includes Velvet EMBOSS and Bowtie2 requiring only the installation of Lambda and PAUDA Installing VirusLand Unzip the VirusLand installation package maintaining the file structure VirusLand v1 0 User s Manual INTRODUCTION This paper presents the documentation of the VirusLand software developed for the taxonomic and functional classification of raw viral sequences within large metagenomic datasets VirusLand has been created as a means to perform metageonomic analysis that is expedient and efficient without sacrificing accuracy and precision It is meant to be simple and straight

File

Contents

Download Pdf Manuals

Related Search

Related Contents