This is the new server for GMOD.org. Please let us know if you notice anything weird while it's getting broken in.

JBrowse Tutorial Programming for Biology 2016

From GMOD
Jump to: navigation, search

This JBrowse tutorial was presented by Scott Cain at the Programming for Biology at Cold Spring Harbor Laboratory in October 2016 using JBrowse 1.12.1.

This tutorial assumes an Ubuntu virtual machine at AWS (ami-47f1b050, in us-east-1), with the security group modified to allow http connections, and with the tutorial bundle zip file, also available on Amazon S3: Prog_for_Bio_2016_JBrowse.zip.

Prerequisites

Prerequisites are installed by JBrowse automatically. A few things may fail to install (like legacy support for wiggle files), but that doesn't matter.

Make sure you can copy/paste from the wiki.

It's also very useful to know how to tab-complete in the shell.

It's probably a good idea to use a browser like Chrome that has the ability to turn off caching while working on this tutorial. To do this in Chrome, with the browser open to the JBrowse page you're working on, select Developer->Javascript Console from the View menu. In the console, click the "gear" icon (settings) and check the box labeled "Disable Cache".

JBrowse Introduction

How and why JBrowse is different from most other web-based genome browsers, including GBrowse.

More detail: paper

JBrowse presentation

Login in to your machine

Obtain the key file (something.pem) and machine IP address from the course website. Then use ssh from the terminal to access the machine:

 ssh -i something.pem ubuntu@ip.add.ress

This will log you in as the ubuntu user, which as full, password-less sudo permissions on the machine.

Setting up JBrowse

Getting JBrowse

  • download the demo bundle from Amazon and unzip it
cd /var/www/html
##curl -O https://s3.amazonaws.com/jbrowse-tutorials/Prog_for_Bio_2016_JBrowse.zip #that's a capital dash"O" not a zero/zed.
cp ~/Prog_for_Bio_2016_JBrowse.zip . ## if we don't need to update the zip file
unzip Prog_for_Bio_2016_JBrowse.zip
cd Prog_for_Bio_2016_JBrowse
unzip JBrowse-1.12.1.zip
mv JBrowse-1.12.1 jbrowse
  • run setup.sh to configure this copy of JBrowse
cd jbrowse
./setup.sh

Starting Point

Visit in web browser:

http://ip.add.re.ss/Prog_for_Bio_2016_JBrowse/jbrowse/index.html

You should see a "Congratulations" page.

Basic Steps

There are four basic steps to setting up an instance of JBrowse:

  1. Load and format reference sequences
  2. Format data for tracks
  3. Configure direct-access tracks
  4. Index feature names

A Short Detour for GFF

GFF (Generic Feature Format) is a very commonly used text format for describing features that exist on sequences. We'll head off to that page to talk about it a bit.

Features from a directory of files

Here, we'll use the Bio::DB::SeqFeature::Store adaptor in "memory" mode to read a directory of files. There are adaptors available for use with many other databases, such as Chado and Bio::DB::GFF.

Config file: pythium-1.conf

{
  "description": "Programming for Biology 2015 P. ultima Example",
  "db_adaptor": "Bio::DB::SeqFeature::Store",
  "db_args" : {
      "-adaptor" : "memory",
      "-dir" : ".."
   },
...

Specify reference sequences

The first script to run is bin/prepare-refseqs.pl; that script is the way you tell JBrowse about what your reference sequences are. Running bin/prepare-refseqs.pl also sets up the "DNA" track.

Run this from within the jbrowse directory (you could run it elsewhere, but you'd have to explicitly specify the location of the data directory on the command line).

cd ~/Sites/jbrowse_demo/Prog_for_Bio_2015_JBrowse/jbrowse
bin/prepare-refseqs.pl --gff ../scf1117875582023.gff

Refresh it in your web browser, you should new see the JBrowse UI and a sequence track, which will show you the DNA base pairs if you zoom in far enough.

Load Feature Data

Next, we'll use biodb-to-json.pl to get feature data out of the database and turn it into JSON data that the web browser can use.

In this case, we have specified all of our track configurations in pythium-1.conf.

...
 
  "TRACK DEFAULTS": {
    "class": "feature"
  },
 
 "tracks": [
    {
      "track": "Genes",
      "key": "Genes",
      "feature": ["mRNA"],
      "autocomplete": "all",
      "class": "transcript",
      "subfeature_classes" : {
            "CDS" : "transcript-CDS",
            "UTR" : "transcript-UTR"
      }
    },
   ...
]

track specifies the track identifier (a unique name for the track, for the software to use). This should be just letters and numbers and - and _ characters; using other characters makes things less convenient.

key specifies a human-friendly name for the track, which can use any characters you want.

feature gives a list of feature types to include in the track.

autocomplete including this setting makes the features in the track searchable.

urltemplate specifies a URL pattern that you can use to link genomic features to specific web pages.

class specifies the CSS class that describes how the feature should look.

For this particular track, I've specified the transcript feature class.

Run the bin/biodb-to-json.pl script with this config file to format this track, and the others in the file:

bin/biodb-to-json.pl --conf ../pythium-1.conf

Refresh JBrowse in your web browser. You should now see a bunch of annotation tracks.

Index feature names

When you generate JSON for a track, if you specify "autocomplete" then a listing of all of the feature names from that track (along with feature locations) will also be generated and used to provide feature searching and autocompletion.

The bin/generate-names.pl script collects those lists of names from all the tracks and combines them into one big tree that the client uses to search.

bin/generate-names.pl -v

Visit in web browser, try typing a feature name, such as maker-scf1117875582023-snap-gene-0.26-mRNA-1. Notice that JBrowse tries to auto-complete what you type.

Features from GFF3 or BED files

We're going to add a couple more tracks that come from a flat file, repeats.gff. To get feature data from flat files into JBrowse, we use flatfile-to-json.pl.

  • We'll add a RepeatMasker track:
bin/flatfile-to-json.pl --trackLabel "repeat masker" \
    --trackType CanvasFeatures \
    --type match:repeatmasker --key RepeatMasker \
    --className generic_parent \
    --subfeatureClasses '{"match_part" : "feature"}' --gff ../repeats.gff
  • And then a RepeatRunner track:
bin/flatfile-to-json.pl --trackLabel "repeat runner" \
    --trackType CanvasFeatures \
    --type protein_match:repeatrunner \
    --key RepeatRunner --className generic_parent \
    --subfeatureClasses '{"match_part" : "feature"}' --gff ../repeats.gff

Visit in web browser; you should see the two new RepeatMasker and RepeatRunner tracks.

BAM alignments

JBrowse can display alignments directly from a BAM file on your web server. Simply place the BAM file in a directory accessible to your web server, and add a snippet of configuration to JBrowse to add the track, similar to:

     {
        "label" : "bam_alignments",
        "key" : "BAM alignments",
        "storeClass" : "JBrowse/Store/SeqFeature/BAM",
        "urlTemplate" : "../../simulated-sorted.bam",
        "type" : "Alignments2"
      }

This can be added by either editing the data/trackList.json file with a text editor, or by running something like this at the command line to inject the track configuration:

echo '{
       "label" : "bam_alignments",
       "key" : "BAM alignments",
       "storeClass" : "JBrowse/Store/SeqFeature/BAM",
       "urlTemplate" : "../../simulated-sorted.bam",
       "type" : "Alignments2"
     }' | bin/add-track-json.pl data/trackList.json

BAM coverage

     {
        "label" : "bam_coverage",
        "key" : "BAM Coverage",
        "storeClass" : "JBrowse/Store/SeqFeature/BAM",
        "urlTemplate" : "../../simulated-sorted.bam",
        "type" : "SNPCoverage"
      }

Quantitative data

BigWig

JBrowse can display quantitative data directly from a BigWig file on your web server. Simply place the BigWig file in a directory accessible to your web server, and add a snippet of configuration to JBrowse to add the track, similar to:

     {
        "label" : "bigwig_bam_coverage",
        "key" : "BigWig - BAM coverage",
        "storeClass" : "BigWig",
        "urlTemplate" : "../../simulated-sorted.bam.coverage.bw",
        "type" : "JBrowse/View/Track/Wiggle/XYPlot",
        "variance_band" : true
      }

This can be added by either editing the data/trackList.json file with a text editor, or by running something like this at the command line to inject the track configuration:

echo ' {
       "label" : "bigwig_bam_coverage",
       "key" : "BigWig - BAM coverage",
       "storeClass" : "BigWig",
       "urlTemplate" : "../../simulated-sorted.bam.coverage.bw",
       "type" : "JBrowse/View/Track/Wiggle/XYPlot",
       "variance_band" : true
     } ' | bin/add-track-json.pl data/trackList.json

Variation Data

VCF tracks

JBrowse can also display VCF variation data directly from a VCF file on your web server that has been compressed with Heng Li's bgzip and tabix. Simply place the .vcf.gz and .vcf.gz.tbi files in a directory accessible to your web server, and add a snippet of configuration to JBrowse to add the track, similar to:

      {
        "label" : "bam_variation",
        "key" : "VCF simulated variation",
        "storeClass" : "JBrowse/Store/SeqFeature/VCFTabix",
        "urlTemplate" : "../../simulated-sorted.vcf.gz",
        "type" : "HTMLVariants"
      }

This can be added by either editing the data/trackList.json file with a text editor, or by running something like this at the command line to inject the track configuration:

echo ' {
       "label" : "bam_variation",
       "key" : "VCF simulated variation",
       "storeClass" : "JBrowse/Store/SeqFeature/VCFTabix",
       "urlTemplate" : "../../simulated-sorted.vcf.gz",
       "type" : "HTMLVariants"
     } ' | bin/add-track-json.pl data/trackList.json


Faceted Track Selection

JBrowse has a very powerful faceted track selector that can be used to search for tracks using metadata associated with them.

The track metadata is kept in a CSV-format file, with any number of columns, and with a "label" column whose contents must correspond to the track labels in the JBrowse configuration.

The demo bundle contains an example trackMetadata.csv file, which can be copied into the data directory for use with this configuration.

cp ../trackMetadata.csv data/

Then a simple faceted track selection configuration might look like:

   "trackSelector": {
       "type": 'Faceted',
   },
   "trackMetadata": {
       "sources": [
          { "type": 'csv', "url": 'data/trackMetadata.csv' }
       ]
   }

Copy the section above and put it in the empty curly braces in the jbrowse_conf.json file in the jbrowse, save it, refresh your browser, and you should now see the faceted track selector activated.

Changing the way tracks look

Tracks can be modified by changing several aspects of how the images are created. While this can be done be done both with HTML and Canvas tracks, this tutorial will focus on Canvas tracks only (the repeat tracks created above).

The configuration for the RepeatMasker track looks like this:

      {
         "key" : "RepeatMasker",
         "trackType" : "CanvasFeatures",
         "storeClass" : "JBrowse/Store/SeqFeature/NCList",
         "urlTemplate" : "tracks/repeat masker/{refseq}/trackData.json",
         "style" : {
            "subfeatureClasses" : {
               "match_part" : "feature"
            },
            "className" : "generic_parent"
         },
         "type" : "CanvasFeatures",
         "compress" : 0,
         "label" : "repeat masker"
      }

Open the data/trackList.json file in your favorite editor and Command-F will open a "find" window; search for "repeatmasker". A simple change we can make is to the color; in the line starting with "style", add:

  "color" : "black",

save the changes and select the RepeatMasker track or reload the browser to see the change. Many attributes of the display can be modified in this way, see the JBrowse Configuration Guide for a list of options.

Making changes based on the data

Much like GBrowse's perl callbacks that can change the track display, in the JBrowse configuration file you can include JavaScript functions to change the way tracks look. For example, in the RepeatMasker track, we can change the color of the glyph depending on what kind of repeat it is (where we happen to know that the type of repeat is encoded in the name). In this example, we leave the glyph black, unless it is a low complexity repeat, where we'll color it red. A function to do that would look like this:

  "color" : "function(feature) { var name = feature.get('Name'); if (name.match('Low_complexity') ) { return 'red'; } return 'black';  }",

When editing the trackList.json file directly in this way, the function has to go all on one line, but if we create an "include file" (not covered here) the function can have carriage returns in it. Replace the simple "color : black" section we just created in the configuration file with the function above, save the file and reload the browser page to see the changes (you might have to mouse around to find a low complexity repeat).

Making links open something else

The default action when you click on a glyph is to open a "floating" window that displays everything JBrowse knows about a feature. If you'd like something else to happen, you have several options (outlined here), including having a different floating window open or executing any JavaScript function you define. For this example, we'll create a link that will Google the repeat's name and open the result in a new window. In the RepeatMasker section of the JBrowse configuration, we'll add a section that looks like this after the style section:

        "onClick" : {
           "iconClass" : "dijitIconDatabase",
           "action" : "newWindow",
           "url" : "http://www.google.com/search?q={name}",
           "label" : "Search for {name} at Google",
           "title" : "function(track,feature,div) { return 'Searching for '+feature.get('name')+' at Google'; }"
        },

If you're having difficulties, the RepeatMasker section of the configuration file should now look something like this:

      {
         "type" : "CanvasFeatures",
         "trackType" : "CanvasFeatures",
         "style" : {
            "color" : "function(feature) { var name = feature.get('Name'); if (name.match('Low_complexity') ) { return 'red'; } return 'black';  }",
            "subfeatureClasses" : {
               "match_part" : "feature"
            },
            "className" : "generic_parent"
         },
         "onClick" : {
           "iconClass" : "dijitIconDatabase",
           "action" : "newWindow",
           "url" : "http://www.google.com/search?q={name}",
           "label" : "Search for {name} at Google",
           "title" : "function(track,feature,div) { return 'Searching for '+feature.get('name')+' at Google'; }"
         },
         "compress" : 0,
         "storeClass" : "JBrowse/Store/SeqFeature/NCList",
         "label" : "repeatmasker",
         "urlTemplate" : "tracks/repeatmasker/{refseq}/trackData.json",
         "key" : "RepeatMasker"
      },

Using Plugins

JBrowse is built with a very flexible and powerful plugin system. The JBrowse developers are working on a plugin registry website, but for the time being, you can look at the source code for what will drive the website in the jbrowse-registry github repo, and in particular, the file that contains the info about the available plugins, plugins.yaml.

For this tutorial, we'll look at a plugin that is shipped with JBrowse but isn't turned on by default. JBrowse plugins are typically stored in the plugins directory, and in 1.12.1's plugin directory there are 6 plugins:

 CategoryUrl
 DebugEvents
 HideTrackLabels
 NeatCanvasFeatures
 NeatHTMLFeatures
 RegexSequenceSearch

and the RegexSequenceSearch plugin is already activated (look under the "Track" menu for it). We will turn on the NeatHTMLFeatures plugin. Right now the Genes track has relatively "boring" glyphs composed of pink rectangles connected by black lines. To turn on NeatHTMLFeatures, just add

   "plugins": [ 
      'NeatHTMLFeatures'
   ],

to the top of data/trackList.json, just inside the first open curly brace. Reload JBrowse and turn on the Genes track if it isn't already and see the wonder of the psuedo-3D gene glyphs.

Note that not all plugins are activated this way: typically, if the plugin modifies the way tracks look, it will go here. If it modifies the way JBrowse *works*, it will go in the jbrowse.conf file.

JBrowse Features

Highlighting interesting things

To highlight a region, you can either right-click on a feature and select 'highlight this', or you can set the highlight explicitly to a certain genomic region by clicking "View -> Set highlight" in the menu bar.

Beginning in JBrowse 1.10.0 you can also highlight a region with the mouse by clicking the highlighter tool (next to the Go button) and clicking and dragging to highlight a region.

Opening local files

JBrowse can display GFF3, BAM, BigWig, and VCF+Tabix files directly from your local machine without the need to transfer any data to the server. Just use the "File -> Open" tool from the menu bar to add tracks using local files.

Combination tracks

Starting in version 1.10.0, users can define tracks that are combinations of the data in other tracks. The operations used to combine these tracks can be set operations (union, intersection, subtraction), arithmetic operations for quantitative tracks (addition, subtraction, multiplication, division), and/or masking operations to just highlight or mask some regions based on data in another track.

To add a combination track, select "File->Add combination track" from the menu bar, and drag existing tracks into the new combination track to start combining them.

Upgrading an Existing JBrowse

If the old JBrowse is 1.3.0 or later, simply move the data directory from the old JBrowse directory into the new JBrowse directory after running the setup.sh script.

Common Problems

  • JSON syntax errors in configuration files (2.x series will stop this madness!)

Other links