Experience With Loading Ngs Data On Standalone Instance Of Galaxy

9.4 years ago by

Penn State

Abhishek: Let talk. This is the area of active current development. We are looking at implementing a universal fastq-like format or supporting multiple formats. Perhaps we should join efforts in ironing out specifications. anton galaxy team Anton Nekrutenko http://nekrut.bx.psu.edu http://galaxyproject.org

ADD COMMENT • link written 9.4 years ago by Anton Nekrutenko ♦ 1.7k

Hello Abhi, Can you clarify the steps you took that produced the behavior? See my comments below. Are you using the Galaxy upload utility to create an item in your history that points to the dataset file on disk? I am not sure what is the rationale A data file that originates from a place external to Galaxy must be uploaded into Galaxy so that the disk file can be placed in the location configured in the Galaxy config file. Also, when data is uploaded to Galaxy ( either to a history or a library ), several database table settings are created that are used by various Galaxy features. They could actually be used straight What do you mean by "the path specified"? Also is there any way to access the You should be able to run any Galaxy tool from the command line as long as you have all of the tool's required binaries in your path. However, running a tool from within Galaxy should generally not be any slower than running it outside of Galaxy, depending, of course, on what you are doing.

ADD REPLY • link written 9.4 years ago by Greg Von Kuster • 840

Hi All @Greg : Please find my comments below. Yes that is precisely correct, I am trying to upload a solexa FASTQ file but on a standalone galaxy installation from my local file system. Thanks for the clarification but I am not sure this will help a lot of people who are interested to install and run galaxy locally mainly for the following reasons. May be it is just local to me. A. We already one instance of data saved on the local file system B. Making another copy via galaxy will eat away a lot of space in long run. C. The time needed to import the files into galaxy space is huge Well what I mean was a way to specify the path of the file/run on the lcoal file system and galaxy could directly pick it up from there rather than uploading it into its own space. Now I understand this might not work based on the way the system was designed. Ok I was under the impression that running from SHELL will eliminate the step of uploading them into galaxy file space. -Abhi

ADD REPLY • link written 9.4 years ago by Abhishek Pratap • 110

Hello Abishek, We are currently in the process of significantly enhancing the current Galaxy upload utilities, and the new version should eliminate the issue you've raised about the time needed to upload large files via HTTP ( not for making an initial copy of the file in the Galaxy environment ). However, it will probably not be ready for release for a few more weeks, so if you can take advantage of Assaf's script in the meantime, that's great. I can't guarantee that all Galaxy features will function correctly if you do this though. Assaf, have you found that using your script breaks anything? Also, if you upload a file to a library rather than a history, multiple users can "import" the library dataset into their history for analysis, but there is only 1 file on disk ( users are pointing to it from their histories ). But uploading a file to a history will create a new copy of the file each time it is uploaded. Greg Von Kuster Galaxy Development Team

ADD REPLY • link written 9.4 years ago by Greg Von Kuster • 840

Hi Greg, Anton and all Just wondering if there has been any progress made on this end. I am sorry I was not able to follow it up on Assaf's suggestion due to other things at work. I did try the latest version of galaxy and looks like the files are still transferred over HTTP before they could be used in the galaxy workspace. Also I would again like to highlight that many labs might want to use the local instance of galaxy and prefer to point to a local path where the file is being stored. That way we will have both the benefits of using a cool GUI and process data stored locally. Let me know if you guys need some feedback or have more questions. I will be happy to discuss them. best, -Abhi

ADD REPLY • link written 9.2 years ago by Abhishek Pratap • 110

If you need a more elaborate solution creating folders per group and user you could take a look at what I have written up here: http://idotamir.blogspot.com/2009/08/adventures-in-galaxy-pt1.html http://idotamir.blogspot.com/2009/09/adventures-in-galaxy-pt2.html The solution given at galaxy-central-importer should just work(tm) and show you what you can do. But just don't merge it into the current version - this will not work. best, ido

ADD REPLY • link written 9.2 years ago by Ido M. Tamir • 280

Hello Abhishek, The Galaxy distribution includes the enhancements to which I previously referred for uploading history files. Uploading files to a history now creates a Galaxy job just like any other tool, and can be run on a cluster node, allowing upload of very large files. The initial pass of this work is also completed for uploading to a Data Library, but this enhancement is still in test, so it should soon be available in the distribution. Do you want to avoid having to import at all (e.g. allow Galaxy to refer to datasets that live in their original locations)? This is not currently possible, but if this is what you are looking for, we can consider some additional options on the current upload form, or possibly a new, separate form. Greg Von Kuster Galaxy Development Team

ADD REPLY • link written 9.2 years ago by Greg Von Kuster • 840

HI Greg Thanks for a quick reply and making some requested changes. However I am not still sure if importing NGS data will help in long run. For Centers generating NGS data which could 2-3 T.B / week depending on no. of sequencers I think importing another copy of raw data into galaxy workspace will be asking for lot of disk space. I understand it is a neat way of doing things as it becomes agnostic of the raw data location but might not be the best way for handling huge data in long run for centers like ours. Please correct me if I am wrong. I think we could also have a simple option without having to import the data and just using it for analysis from the current location, also storing results at the same location. That way in future even if the data set is moved analysis also stays with it. Let me know what you feel. I will be happy to know if there are any other smart reasons of importing the data in galaxy workspace just for curiosity sake. Thanks, -Abhi

ADD REPLY • link written 9.2 years ago by Abhishek Pratap • 110

Dear all, to echo what Abhi said: we are also currently looking of ways to automatically import data sets (libraries) into Galaxy without having to manually trigger the import via the administration interface, and ideally while keeping the data in the original place. The idea here is to have multiple tools all point at the original 'source data' without having to replicate terabytes of data. Not quite sure how feasible this is in practice, but it certainly would be incredibly helpful. Best, Oliver -- Research Associate Department of Biostatistics Associate Director Bioinformatics Core Harvard School of Public Health Skype: ohofmann Phone: +1 (617) 365 0984

ADD REPLY • link written 9.2 years ago by Oliver Hofmann • 50

Dear Oliver, It is quite simple. If you have 7 minutes do the following from the command line: hg clone http://bitbucket.org/ido/galaxy-central-importer gci cd gci hg update -C importer sh setup.sh #sets up basic things ./run.sh #end with ctrl-c after startup completes (serving on ....) ./importer.sh #starts the importer wait for "done importing" ./run.sh then open your browser at localhost:8080/galaxy You can log in with 3 different usernames/passwords: name1@/123456 #can only see group1 libs name2@123456 #can only see group2 libs tamir@/123456 #can see all libs Of course only the file paths get inserted into the db. more on: http://idotamir.blogspot.com best wishes, ido

ADD REPLY • link written 9.2 years ago by Ido M. Tamir • 280

Change set 2812 will be included in a release to the distribution today - here are details of a new option that we're hoping will provide what is needed for most labs. Add a new option, 'allow_library_path_paste' that adds a new upload page ("Upload files from file system paths") to the admin-side library upload pages. This form contains a textarea that allows Galaxy admins to paste any number of file system paths (files or directories) from which Galaxy will import library datasets, saving the directory structure (if desired). Since such ability allows admins access to any file on the Galaxy server which is readable by Galaxy's system user, this option is disabled by default, and system administrators should take care in assigning Galaxy administrators when this feature is enabled. Controls on what files are accessible to this tool based on ownership or other properties can be added at a later date if there is sufficient interest for such features. This commit also includes a checkbox on the "Upload directory of files" page (as well as the new "Upload files from file system paths" page above) that will prevent Galaxy from copying data to its files directory (by default, 'database/files/'). This is useful for large library datasets that live in their own managed locations on the file system, this will prevent the existence of duplicate copies of datasets (but means administrators must take care to manage data - moving or removing the data from its Galaxy-external location will render these datasets invalid within Galaxy). One unique feature to be aware of: when using the "Copy data into Galaxy?" checkbox on the "Upload directory of files" page, any symbolic links encountered in the chosen import directory will be made absolute and dereferenced ONCE. This allows administrators to link large datasets to the import directory, rather than having to make full copies, while being able to delete such links after importing. Only the first symlink (the one in the import directory itself) is dereferenced; all others remain. See the following for an example: library_import_dir = /galaxy/import % ls -lR /galaxy/import /galaxy/import: total 6 drwxr-xr-x 2 nate nate 512 Oct 1 11:31 link/ /galaxy/import/link: total 10 lrwxrwxrwx 1 nate nate 71 Oct 1 10:38 1.bed -> ../../../home/nate/galaxy/test-data/1.bed lrwxrwxrwx 1 nate nate 60 Oct 1 10:38 2.bed -> /home/nate/galaxy/test-data/2.bed lrwxrwxrwx 1 nate nate 11 Oct 1 10:38 3.bed -> ../../3.bed lrwxrwxrwx 1 nate nate 35 Oct 1 11:30 4.bed -> ../../galaxy_symlink/test-data/4.bed lrwxrwxrwx 1 nate nate 41 Oct 1 11:31 5.bed -> /galaxy/galaxy_symlink/test-data/5.bed % ls -l /galaxy/3.bed lrwxrwxrwx 1 nate nate 60 Oct 1 10:39 /galaxy/3.bed -> /home/nate/galaxy/test-data/3.bed % ls -l /galaxy/galaxy_symlink lrwxrwxrwx 1 nate nate 44 Oct 1 11:30 /galaxy/galaxy_symlink -> /home/nate/galaxy/ In this example, 1.bed is a relative symbolic link to the real 1.bed. 2.bed is an absolute symlink to the real 2.bed. 3.bed is a relative symlink to ../../3.bed, aka /galaxy/3.bed, which itself is a symlink to the real 3.bed. 4.bed is a relative symlink which follows another symlink (/galaxy/galaxy_symlink) to the real 4.bed. 5.bed is an absolute symlink in the same fashion as 4.bed If the 'link' server directory is chosen on the "Upload directory of files" page, and "Copy data into Galaxy?" is checked "No", the following files will be referenced by Galaxy: /home/nate/galaxy/test-data/1.bed /home/nate/galaxy/test-data/2.bed /galaxy/3.bed /galaxy/galaxy_symlink/test-data/4.bed /galaxy/galaxy_symlink/test-data/5.bed The Galaxy administrator may now safely delete /galaxy/import/link, but should take care not to remove the referenced symbolic links (/galaxy/3.bed, /galaxy/galaxy_symlink). Not all symbolic links are dereferenced because it is assumed that if an administrator links to a path in the import directory which itself is (or contains) links, that is the preferred path for accessing the data.

ADD REPLY • link written 9.2 years ago by Greg Von Kuster • 840

Hi Greg Many thanks for accommodating our requests in quick time. I will be testing it right away. Have a good weekend. Cheers, -Abhi

ADD REPLY • link written 9.2 years ago by Abhishek Pratap • 110

Hi Greg I have updated my galaxy rep to changeset 2825. I dont see the checkbox on the "Upload File" page. Am I missing something ? Thanks, -Abhi

ADD REPLY • link written 9.2 years ago by Abhishek Pratap • 110

Hello Abhishek, Add this to your universe_wsgi.ini file: allow_library_path_paste = True Then, clicking the down-arrow on the upload form Create new data library datasets ▼ will give you 4 options, 1 of which is: Upload files from file system paths Greg Von Kuster Galaxy Development Team

ADD REPLY • link written 9.2 years ago by Greg Von Kuster • 840

Hi Greg Unfortunately it is not working for me. I made sure I cleared my browser cache before re-viewing it. I have set the option as suggested by you in the universe_wsgi.ini file. -Abhi

ADD REPLY • link written 9.2 years ago by Abhishek Pratap • 110

Please type the following in your galaxy install directory, and let me know what you get: hg heads Thanks

ADD REPLY • link written 9.2 years ago by Greg Von Kuster • 840

Here it is : changeset: 2824:d97f4e86be45 tag: tip parent: 2823:2c0c81150dbd parent: 2821:04a753865407 user: Anton Nekrutenko <anton@bx.psu.edu> date: Fri Oct 02 11:02:14 2009 -0400 summary: merge -Abhi

ADD REPLY • link written 9.2 years ago by Abhishek Pratap • 110

Ok, you have the change set. Perhaps you are trying to upload into a history rather than a Data Library? This new feature is only available for uploading into a Data Library. Can you confirm? I'm not sure what else could be causing you to not see the 4 options on the drop-down menu for uploading dataset into a Data Library. Greg

ADD REPLY • link written 9.2 years ago by Greg Von Kuster • 840

I think that could be it. I am using the "Upload File " option under the "Get Data" from the left hand menu. Sorry for the confusion if I am doing something that I am not supposed to. How do we upload data to data library ? -Abhi

ADD REPLY • link written 9.2 years ago by Abhishek Pratap • 110

Set yourself up as a Galaxy admin, in universe_wsgi.ini, add your galaxy account ( email ) to the following: # this should be a comma-separated list of valid Galaxy users admin_users = you@youremail.edu You'll see an Admin link in the top Galaxy menu bar when you restart your Galaxy sever after making this change. Click on it, and you be presented with teh Galaxy Admin UI. Look for the "Manage data libraries" link in th left panel, create a new data library, and you can upload datasets to using 1 of 4 options, including this latest option just introduced. Greg Von Kuster Galaxy Development Team

ADD REPLY • link written 9.2 years ago by Greg Von Kuster • 840

Hi! One more question according change set 2812: Add a new option, 'allow_library_path_paste' that adds a new upload page ("Upload files from file system paths") to the admin-side library upload pages. I am not sure where to add this option- i added to universe_wsgi.ini: allow_library_path_paste = True allow_library_path_paste = /somepath/ both didnt work- how can i enable the "allow_library_path_paste" option? thanks! mat Greg Von Kuster schrieb: -- Matthias Dodt Scientific Programmer at Bioinformaitcs platform AG Dieterich Berlin Institute for Medical Systems Biology at the Max-Delbrueck-Center for Molecular Medicine Robert-Roessle-Strasse 10, 13125 Berlin, Germany fon: +49 30 9406 4261 email: matthias.dodt@mdc-berlin.de

ADD REPLY • link written 9.1 years ago by Matthias Dodt • 110

Hi Guys! Thanks for the reply - I wrote a simple .xml which allows the import of any file which is accessible by the galaxy process (read permission) and is mounted in the current file system of the galaxy server. in the web-interface you have to type in the absolute path+filename manually- It copies the file into galaxy (unix cp used). The uploaded file is assumed to be a text file and has to be set manually to the correct file format- @nate: Thanks! ill have a look at that too- Greetings, mat <tool id="importer_1" name="Cluster file importer"> <description>copies files from import directory into galaxy</description> <command> cp $source $target 2> $log_report </command> <inputs> <param name="source" type="text" label="Absolute path+filename to file on cluster" optional="false" size="60"/> </inputs> <outputs> <data name="target" format="text" label="Imported file"/> <data format="text" name="log_report" label="Detailed log report from importer"/> </outputs> <help> **What it does** This tool imports a file from the cluster into galaxy. The file will be copied into the galaxy environment, the original file remains on the cluster. **!Important advice!** Please make sure the uploaded file is set to the correct file format! Otherwise tools cannot access this file. This can be done by editing the properties of the uploaded file (pen symbol in the history). The default assumed file format is -text- which might be incorrect it most cases. Please keep although in mind that disk space on the galaxy server is limited. **Example** /home/mat/myfile.fa **Support** Feel free to contact us if you cant upload your files into galaxy. </help> </tool>

ADD REPLY • link written 9.2 years ago by Matthias Dodt • 110

Good Afternoon Anton: I've written a fastq parser here. As I did that I discovered that existing fastq formats are ambiguous and not really identifiable from their data. Are you thinking of making a new fastq specification ? --Hiram - UCSC genome browser

ADD REPLY • link written 9.4 years ago by Hiram Clawson • 260

Hiram: Can you share the code? Which baseQ scaling are you using (Sanger| Solexa)? SOliD support? Thanks, anton Anton Nekrutenko http://nekrut.bx.psu.edu http://galaxyproject.org

ADD REPLY • link written 9.4 years ago by Anton Nekrutenko ♦ 1.7k

kent source tree directories: src/utils/fastqToFa/ and src/utils/faToFastq/ I've never used the faToFastq program. The fastqToFa takes arguments to decide the baseQ score scheme. See usage message below. I haven't tested the -solexa scoring method option. --Hiram fastqToFa - Convert from fastq to fasta format. usage: fastqToFa [options] in.fastq out.fa options: -nameVerify='string' - for multi-line fastq files, 'string' must match somewhere in the sequence names in order to correctly identify the next sequence block (e.g.: -nameVerify='Supercontig_') -qual=file.qual.fa - output quality scores to specifed file (default: quality scores are ignored) -qualSizes=qual.sizes - write sizes file for the quality scores -noErrors - warn only on problems, do not error out (specify -verbose=3 to see warnings -solexa - use Solexa/Illumina quality score algorithm (instead of Phread quality) -verbose=2 - set warning level to get some stats output during processing

ADD REPLY • link written 9.4 years ago by Hiram Clawson • 260

Dr. Nekrutenko, I've found the fq_all2std.pl script distributed with Maq to be very helpful for converting file formats to a standardized Fastq. Because it's so small (10K) I'll attach it to this email for anyone interested. Also, I'll copy its usage notes here so you can see the formats it converts: Usage: fq_all2std.pl <command> <in.txt> Command: scarf2std Convert SCARF format to the standard/Sanger FASTQ fqint2std Convert FASTQ-int format to the standard/Sanger F ASTQ sol2std Convert Solexa/Illumina FASTQ to the standard FAS TQ std2sol Convert standard FASTQ to Solexa/Illumina FASTQ ( simplified) fa2std Convert FASTA to the standard FASTQ seqprb2std Convert .seq and .prb files to the standard FASTQ fq2fa Convert various FASTQ-like format to FASTA export2sol Convert Solexa export format to Solexa FASTQ export2std Convert Solexa export format to Sanger FASTQ csfa2std Convert AB SOLiD read format to Sanger FASTQ std2qual Convert standard FASTQ to .seq+.qual instruction Explanation to different format example Show examples of various formats Note: Read/quality sequences MUST be presented in one line. John Obenauer -- John Obenauer, Ph.D. Bioinformatics Group Leader Information Sciences Department St. Jude Children's Research Hospital 262 Danny Thomas Place Mail Stop 312 Memphis, TN 38105 (901) 595-3188 john.obenauer@stjude.org Email Disclaimer: www.stjude.org/emaildisclaimer

ADD REPLY • link written 9.4 years ago by John Obenauer • 10

Similar posts • Search »