Metaxa User's Guide 1.1.2 - Microbiology, Metagenomics and
Contents
1. else off F Selects whether to use HMMER s heuristic filtering Off F by default Turning this setting on will increase speed but decrease precision Uses megablast for classification for better speed but less accuracy Off F by default If on Metaxa outputs a summary of results File suffix is ssummmary txt On T by default If on Metaxa outputs graphical text representations of where in each sequence the conserved domains were found File suffix is graph On T by default If on FASTA formatted files containing the extracted SSU sequences are written One file for each origin is written plus an extraction file containing all SSUs identified in the first analysis step On T by default If on Metaxa saves table format output of results separately for HMMER and BLAST output Note that neither of these outputs is the actual output of the respective program To get these file use the save_raw T see below Off F by default not_found T or F If on Metaxa outputs a list of entries that do not seem to be SSU sequences File suffix is _not_found txt Off F by default align a all u Outputs alignments of BLAST matches to each query in all a uncertain uncertain n none u or no n cases Requires MAFFT to be installed Default is to output alignments in uncertain cases u truncate T or F Removes ends of SSU sequences if they are outside of the SSU region If off2. microbiology se software metaxa in order to download the Metaxa package Download it to your preferred directory Unpack the downloaded tarball with tar xvfz metaxa tar gz A directory called Metaxa will be created You will see the following files and directories inside it metaxa metaxa_x metaxa_c install_metaxa the metaxa_db directory containing the Hidden Markov Models and a BLAST database the user s guide the README txt file the license txt file as well as test input files Enter the directory and type install_metaxa Press enter and follow the on screen instructions You will be prompted for whether you have superuser privileges and where you want Metaxa to be installed If Metaxa is successfully installed you should see its help message when typing the command metaxa help 2 Usage and commands For the very impatient only follow the brief installation instructions in the file README txt To check for SSU rRNA sequences in the file test fasta you would then type metaxa i test fasta o test on the command line For all other users Metaxa accepts input in the FASTA format As it pre processes the input sequences it is possible to input both aligned and unaligned FASTA files containing both DNA and RNA sequences By default Metaxa outputs ten files one summary file of the entire run one more detailed table of results one graphical representation of hits one FASTA file of all identified SSU sequen
3. bacterial eukaryal mitochondrial and chloroplast origin are utilised To accommodate for the fact that there are both 16S and 12S mitochondrial SSU sequences the latter chiefly found in animals two distinct sets of profiles are used to accurately detect both of these categories The archaeal bacterial and eukaryote profile sets are taken from the V Xtractor 2 0 software while the chloroplast and mitochondrial 12S and 16S sets are newly generated following the same procedure as for the V Xtractor profiles To avoid false positive matches Metaxa by default requires at least two such conserved domains to be found on a query sequence This criterion brings down the false positive rate to about 0 0001 As several of the conserved domains are closely similar in e g bacteria and chloroplasts the initial classification made by the HMMER basced step will not be perfect Thus the results of the extraction are sent to a BLAST based classification step where each SSU sequence is matched to a manually inspected database of archaeal 16S bacterial 16S eukaryal 18S chloroplast 16S and mitochondrial 12S and 16S sequences Each possible origin is assigned points according to the origin of the BLAST matches as well as the origin predicted by HMMER By default the scoring system gives 5 points to the origin predicted by the HMMER based extraction The origin of the best BLAST match to the sequence is also given 5 points and the origins of the subsequent B
4. is part of a pipeline where input files with the same name could cause overwriting of important data Off F by default Sequence selection options t b bacteria a archaea e eukaryota m mitochondrial c chloroplast A all E value S value N value M value H value selection_priority sum domains eval score search_eval value Set of profiles to use for the search comma separated Accepts any list of sets e g bacteria chloroplast m c or eukaryota Can be used to restrict the search to only a few SSU types to save time if one or more of the origins are not relevant to the dataset under study Default is to use all the all option Domain E value cutoff a sequence must obtain in the HMMER based step to be included in the output Default 1 Domain score cutoff for a sequence must obtain in the HMMER based step to be included in the output Default 12 The minimum number of domains that must match a sequence for it to be included in the output Setting the value lower than two will increase the number of false positives while increasing it above two will decrease Metaxa s detection abilities on fragmentary data Default 2 Number of top BLAST matches that should be considered in classification Default 5 The number of points that the predicted origin of the Metaxa Extractor is given Default is the same as the number of sequences used for cl
5. the name lt query identifier gt aligned fasta Chimeric sequences If the option allow_reorder is turned off Metaxa will save an additional FASTA file containing sequences that are suspected to be chimeric These are sequences with domains located in the wrong order This is useful on full length or near full length data sets but should not be used on short reads as it could increase the number of false negatives when run on short sequences Raw data If the option to save all raw data is turned on Metaxa will save all data from the pre processing HMMER search BLAST search as well as a file of raw statistics into a directory with the suffix _metaxa_raw_output 4 Algorithm and implementation The main design goal for Metaxa is to achieve fast and accurate extraction of SSU sequences in large data sets without introducing a large number of false positives To be able to reach a high speed Metaxa relies on the HMMER3 software which allows for extremely fast comparisons of HMM profiles to a sequence set HMMER is used to extract a subset of the input sequences that is subsequently analysed for origin Thus the program does not have to consider a large number of non SSU sequences that would slow down the classification process To achieve high detection accuracy Metaxa uses multiple HMM profiles representing conserved domains in the SSU sequence In addition separate sets of HMM profiles for SSU sequences of archaeal
6. the whole input sequence is saved On T by default guess_species T or F Writes a species guess based on the BLAST matches to the FASTA definition line This guess can be pretty far off Off F by default silent T or F Suppresses printing of progress info to screen Off F by default graph_scale value Sets the scale of the graphical output If the provided value is zero a percentage view is shown Default is 0 Save_raw T or F Saves all raw data for searches etc instead of removing it when finished Saves data to a directory with the suffix _metaxa_raw_output Off F by default Information options h Displays the help message help Displays the help message bugs Displays the bug fixes and known bugs in this version of Metaxa license Displays licensing information 3 Output files Metaxa outputs a number of files depending on what is selected by the user see Usage and Commands above By default seven FASTA files a table of extraction results a file containing graphical representation of putative SSU sequences and a summary file is written In addition tables of BLAST and HMMER results lists of non SSU entries and sequence alignments can be written on request by the user There is also an option to preserve all the intermediate data generated by the HMMER and BLAST searches FASTA output Metaxa generates one FASTA file for each origin archaea bacteria eukaryota chloroplast and mitoch
7. LAST matches is given scores decreasing by one for each match When all BLAST matches have been analysed the score for each origin is summed up and the sequence is assigned to the origin with the highest score If the origin of the final classification does not agree with the predicted origin from the HMMER based step the sequence classification is marked as uncertain by applying a to the end of the definition line The sequence is also marked as uncertain if the difference between the scores of the two most likely origins is smaller than the number of sequences of analysed BLAST matches by default 5 The second classification step makes Metaxa very accurate even on fragmentary sequences by ensuring that two independent methods agree on the predictions made By applying stringent criteria in the extraction step the software is still very robust with respect to false positives and also reasonably fast even on large metagenomic data sets Its performance is slower on large PCR libraries however as more of the sequences will represent the SSU and hence need to be classified in the second step While Metaxa s default settings should be usable in most situations you should consider if they suitable for your purposes and for your data set If the data set is small this can be done by running the software multiple times on the data with different settings and analyse the outcome On larger data sets it might be more feasible to only
8. User s guide Manual for Metaxa 1 1 2 This is a guide to install and use the software utility Metaxa The software is written for Unix like platforms and should work on nearly all Linux based systems as well as MacOS X Contents of this manual 1 Detailed installation instructions Usage and commands Output files Algorithm and implementation Running Metaxa s analysis steps separately Undocumented features Pals SOR a AS 209s IS License information 1 Detailed installation instructions The README txt file bundled with the script provides a quick installation guide In order to install certain packages you might need to have superuser privileges For installation on Mac you will have to install the Apple Xcode package available on your MacOS X System DVD in order to be able to compile programs Please talk to your system administrator if you feel unsure about these steps Note that the packages are mandatory and that you should not proceed unless these criteria are fulfilled If you don t have superuser privileges on your machine Create a directory within your user directory e g home user bin and to store all required binaries there By adding this directory to your PATH any software placed in the directory will behave as if installed for all users using superuser privileges If you use the bash shell you can add a bin directory to your PATH by adding the line export PATH PATH HOME bin to the file profil
9. a file with the suffix summary txt In this file the statistics of the run is collected as are the starting and ending times for the run Also lists of the identifiers of extracted SSU sequences are written to this file one list for each origin The first section of the file shows the data from the extraction step The second section is associated with the second classification step After the second section the lists of entries of different origins are found An example of parts of a summary file is shown below Metaxa run started at Mon Mar 14 10 07 52 2011 Number of sequences in input file 100 Sequences detected as SSU rRNA by Metaxa 100 On main strand 91 On complementary strand 9 SSU sequences by preliminary origin Archaea 0 Bacteria 0 Eukaryota 0 Chloroplast 100 Mitochondria 0 Other 0 Number of SSU rRNA sequences to be classified by Metaxa 100 Number of SSU rRNA having at least one database match 100 Number of SSU rRNA successfully classified by Metaxa 100 Number of uncertain classifications of SSU rRNA sequences 0 Total number of classifications made by Metaxa 100 Number of SSU rRNA sequences assigned to each origin Archaea 0 Bacteria 0 Eukaryota 0 Chloroplast 100 Mitochondria 0 Uncertain 0 Sequences of chloroplast origin 16S Acorus_americanus_AcamCr001 Aethionema_cordifolium_AecoCr001 Welwitschia_mirabilis Wemic_r001 Zea_mays_ZemaCr113 Sequences of mitochondrial origin 12S a
10. assification M option above which is set to 5 by default Determines what will be of highest priority when assessing the origin of the sequence Options are sum which sums the scores for each profile match and divides the sum by the number of profiles of the given type domains which uses the number of domains retrieved of a given type eval which uses the average E value of the found hits score which uses the average score of the found hits Default is to use sum sum of scores The actual E value cutoff used in the HMMER search High numbers may slow down the process Should never be set to a lower value than the E option Cannot be used in combination with the search_score option search_score value blast_eval value blast_score value blast_wordsize value allow_single_domain e value score or F allow_reorder T or F complement T or F cpu value multi_thread T or F heuristics T or F megablast T or F Output options summary T or F graphical T or F fasta T or F table T or F Default is 10 The score cutoff used in the HMMER search Low numbers may slow down the process Should never be set to a higher number than the S option Cannot be used in combination with the search_eval option Default is to used E value cutoff see search_eval above not score The E value cutoff used in the BLAST search High numbers may slo
11. ble output is turned on Metaxa will save statistics of every BLAST match that the sequence in question produces against the database to in a file with the suffix blast table This file consists of tab separated columns containing information on the matches found one BLAST match per line The contents of the columns from left to right are explained in this table Column Description Query ID The identifier of the query sequence Subject ID The identifier of the matching database sequence Score The score this match has obtained in the classification system Species The species name of the database system Score The BLAST score of the match E value The E value of the match as reported by BLAST Each new query is indicated by a comment line e g Query AATT01000235 146421 147977 E List of non SSU sequences If not found output is turned on Metaxa will write a list of sequences for which no conserved SSU regions could be found to a file with the suffix _not_found txt The file contains only the identifiers of the non SSU sequences Sequence alignments By default Metaxa saves alignments of sequences of uncertain origin to a directory with the suffix _alignments The user may specify to instead align all SSU sequences by using the align all option note that this would increase the runtime significantly The five best BLAST matches are aligned to the query sequence and saved to an aligned FASTA file with
12. ces and one FASTA file for each of the six possible origins To list all the available options for Metaxa type metaxa help You can use the test fasta file that comes bundled with the software for a test run This file contains 50 randomly selected SSU entries ten of each origin as well as 10 non SSU sequences In the simplest case Metaxa is run by metaxa i input_file o output Below is a listing of all options Metaxa accepts Boolean options can be turned on with T true or 1 and off using F false or 0 Main options i file o file p directory d database date T or F Nucleotide FASTA input file to investigate Metaxa accepts both aligned and unaligned FASTA If no input is specified Metaxa will read sequences from standard input which means that FASTA sequence can be piped into Metaxa Base for the file names of the output files Suffixes will be added automatically Defaults to metaxa_out A path to a directory containing HMM profile collections representing SSU rRNA conserved regions By default Metaxa assumes to find the databases in the metaxa_db directory located in the same directory as Metaxa itself The BLAST database used for classification By default Metaxa assumes to find the databases in the metaxa_db directory located in the same directory as Metaxa itself Adds a date and time stamp to the output file This can be useful e g if Metaxa
13. e in your home directory The process of adding items to one s PATH varies among systems and shells Close the terminal and open a new one for this change to take effect Perl needs to be installed on the computer Most Unix based systems including Linux and MacOS X have Perl pre installed You can check this by opening a command line terminal and type perl v In case Perl is not installed you have to download http www perl org and compile the program Download and install HMMER version 3 http hmmer janelia org software The current version of Metaxa relies on HMMER version 3 Metaxa will not work with earlier versions of HMMER Download the HMMER package source code to your preferred directory such as home user Open a command line terminal move into the directory with cd home user and unpack the tarball with tar xvfz hmmer 3 0 tar gz Now you must compile HMMER from source files To compile it from source enter the new directory and follow the installation instructions in the file INSTALL If you have trouble compiling HMMER you can try to use the pre compiled binaries available at the HMMER home page After download and unpacking of the tarball the binaries are located in the binaries directory contained within the newly created HMMER directory Move into the binaries directory and move all of its contained files into your preferred bin directory usually either usr local bin or your own bin directory h
14. himeric if the sequence was marked as a potential chimera Empty if not Sequences will only be marked as chimeric if the allow_reorder option is turned off Note that this is not a robust measure against chimeras of all kinds Specific origin A collection of information of all possible origins for the given query Each information entry is a space separated list containing the origin type the number of domains of that type the average E value and the average score e g N 4 8 2e 11 43 475 Extraction results table If table output is turned on Metaxa will save statistics of every profile set that the sequence in question matches to in a file with the suffix hmmer table This file consists of tab separated columns containing information on the SSU sequence found The contents of the columns from left to right are explained in this table Column Description ID The identifier of the query sequence Length The length of the query sequence List of hits Each new column contains information of a profile match Each column is organised as follows lt starting position gt lt ending position gt lt name of matching profile gt lt score gt lt E value gt As in the graphical output file the table file is divided into sections Each section represents one group of sequences and begins with the line X matches on main strand and ends with a line of asterisks Classification results table If ta
15. mally not overlapping so this could be an indication of a bad input sequence The line of asterisks indicates the end of one set of matches Note that the graph should be viewed with a non proportional font such as Courier if loaded into e g Word Extraction results table The full results of the Metaxa extraction is saved to a file with the suffix extraction results This file consists of tab separated columns containing various information on each SSU sequence found The file can be easily imported into programs such as Excel The contents of the columns from left to right are explained in this table Column Description ID The identifier of the query sequence Length The length of the query sequence Origin A one letter abbreviation of the sequence origin A archaeal B bacterial C chloroplast E eukaryote M mitochondrial 16S N mitochondrial 12S Strand A zero 0 if the SSU was found on the main strand a one 1 if it was found on the complementary strand Domains The number of conserved domains for the most likely origin that was found in the sequence Average E value The average E value for these domains Average score The average score for these domains Start The starting position of the first domain End The ending position of the last domain First domain The domain that is located first on the sequence Last domain The domain that is located last on the sequence Chimera The word C
16. nd 16S Metaxa run finished at Mon Mar 14 10 08 42 2011 Graphical representations Metaxa writes graphical ASCII representations of where in each sequence the various conserved regions were found to a text file with the suffix graph Separate graphs are written for each origin and strand which means that each sequence entry may be present more than once in this file if it have matches to HMM profiles from more than one origin This makes it possible to manually inspect how Metaxa has evaluated each sequence The graphical representations look like this B matches on main strand gt gt id 454_ 30 gi 50402825 gb AY687385 1 403 bp kkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkxk The first row shows the type of the entries below as well as the strand they are found on Each entry begins with the characters gt gt followed by the sequence identifier and its length Below the identifier row the sequence graph is shown By default all sequences are scaled so that they are of equal length and the domains are placed according to their relatwe position in the sequence The characters that are used in the graphical representations are explained in the table below Feature Description Part of the sequence without any conserved domain variable region Vil Start of a conserved domain Continuation of a conserved domain gt Indicates that one conserved domain goes into the next Domains are nor
17. ome user bin The HMMER package should now be installed on your computer you can check this by typing hmmscan h in the terminal and press enter you should now see HMMER output Download and install the BLAST package ftp ftp ncbi nlm nih gov blast executables release LATEST for sequence similarity searches The current version of Metaxa relies on BLAST not BLAST and was written with version 2 2 24 in mind It should work with any 2 2 version of BLAST Download the BLAST package for your operating system to your preferred directory Open a command line terminal move into the directory with cd home user and unpack the tarball with tar xvfz blast 2 2 24 platform tar gz Move into the bin directory inside the newly created BLAST directory and move all of its contained files into your preferred bin directory Alternatively you can add the BLAST bin directory to your PATH The BLAST package should now be installed on your computer you can check this by typing blastall in the terminal and press enter you should now see the listing of BLAST options Download and install the MAFFT http mafft cbre jp alignment software for multiple alignment The current version of Metaxa relies on MAFFT version 6 MAFFT is not critical for Metaxa s core functions but is used for automatically creating alignments of uncertain sequences Instructions for installing MAFFT are available on the MAFFT download page Go to http
18. ondria one file containing sequences of uncertain origin and one file with all SSU sequences identified and extracted in the first step Sequences in these files are marked according to their origin Sequences whose origin Metaxa could not establish with certainty but for which enough data were available to allow a qualified guess as to the origin of the sequences are marked with a character at the end of the definition line A certain sequence may look like this gt gi 117927211 Bacterial 16S SSU rRNA GTTTGATCCTGGCTCAGGACGAACGCTGGCGGCGTGCTTAACACATGCAAGTCGAGCGGA Note that Metaxa has added the type of the SSU sequence Bacterial 16S SSU rRNA to the definition line in the example above An uncertain sequence could look like this gt AABL01000014 4508 5931 Putative Chloroplast 16S SSU rRNA GAACGCTAGAAATATACATTACACATGCAAATTTATGATAATATCATAGTGAATAGGTGA The extraction file contains all sequences identified as SSUs by metaxa_x the first step of the analysis The sequence entries in that file contain information on what domains that were found and what origin that is most likely base on the profile search An example is shown below gt A16379 1 1496 B Predicted Bacterial 16S SSU rRNA 1447 bp From domain V1l to V9r on main strand Found domains V11 V21 V2r V31 V3r V4l V4r V51 V5r V6l V71 V81 V8r V91 V9Or CAGGCTTAACACATGCAAGTCGAACGGTAGCACGAAGGACTTGCTCCTTGGGTGACGAGT Summary A summary of the Metaxa run is written to
19. run Metaxa on a sub set of the sequences for testing The graphical output is very useful for determining whether Metaxa performs as desired on the data as the positions of the found conserved domains can be easily investigated If domains are missing the criteria might be set to be too stringent If they are not in sequential order from V11 to V9r that might be an indication that there is something wrong with the input sequences The HMMER program hmmsearch used by Metaxa normally uses heuristic filters to increase the search speed Metaxa runs hmmsearch with the max option in order to turn off all heuristic filters This increases detection power at the cost of speed However the time requirement of the HMMER search is generally not an issue with Metaxa while accuracy is and thus the heuristic filters are not used 5 Running Metaxa s analysis steps separately Metaxa s analysis procedure is divided into two steps the extraction and the classification These two steps are normally run in sequence by running the metaxa command However they can also be run separately if the user wishes To run the extraction step independently use the metaxa_x command This command takes a subset of the metaxa options other options will be ignored To see the available options for the metaxa_x command type metaxa_x help on the command line To run the classification step on a set of SSU sequences use the command metaxa_c The option
20. s for metaxa_c can be seen by typing metaxa_c help on the command line Note that the output files obtained when running each step separately will be slightly different than obtained through running the entire Metaxa pipeline 6 Undocumented features Metaxa has two undocumented options that can be activated but they are considered experimental and should be used with caution One allows you to use pre calculated hmmscan results and feed into Metaxa the other allows using a set of additional HMM profiles for the SSU extraction Undocumented options hmmscan file If the hmmscan has already been performed this option can be used as the base for the hmmscan output files and the hmmscan step will be skipped Overrides the o option while a DNA FASTA input file containing the sequences used for the hmmscan must still be supplied This feature is pretty experimental and was used during early evaluation of Metaxa Use it only with caution t o other It is possible to supply an additional set of HMM profiles in an O hmm file within the HMMs directory This custom set can be any type of profiles but the profiles must be named according to the convention in the other HMM files beginning with V1I and ending with V9r 7 License information This program is free software you can redistribute it and or modify it under the terms of the GNU General Public License as published by the Free Software Foundation either
21. version 3 of the License or at your option any later version This program is distributed in the hope that it will be useful but WITHOUT ANY WARRANTY without even the implied warranty of MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE See the GNU General Public License for more details You should have received a copy of the GNU General Public License along with this program in a file called license txt If not see http www gnu org licenses Copyright C 2011 2013 Johan Bengtsson Palme et al
22. w down the process Cannot be used in combination with the blast_score option Default is 1e 15 The score cutoff used in the BLAST search Low numbers may slow down the process Cannot be used in combination with the blast_eval option Default is to use E value cutoff see blast_eval above not score The word size used for the BLAST based classification Lower numbers will slow down the process significantly while higher numbers may potentially decrease classification accuracy Default is 14 Allow inclusion of sequences that only find a single domain given that they meet the more stringent E value and score thresholds specified By default single domains are allowed with E value cutoff 1e 10 and score cutoff 0 1e 10 0 Allows profiles not to be in the expected order 1 9 on the extracted sequences If turned off a file of potential chimeric sequences with profile matches in the wrong order is written allowing for rudimentary chimera detection This can be used on full length sequences On fragmented sequences however there is a risk of missing true positives increases if this option is turned off On T by default If on Metaxa checks both DNA strands for matches to HMM profiles On T by default The number of CPU threads to use Metaxa performs significantly faster using more CPUs Default is 1 Multi thread the HMMER search On T by default if the number of CPUs is larger than one cpu option gt 1
Download Pdf Manuals
Related Search