Introduction

The following instructions assume you are not going to make a new genome version directory and that all you need to do is update refGene, all_mrna and all_est files.

IGB QuickLoad is configured so that a set of foundation gene annotations (gene models) load into IGB as soon as the user selects the corresponding genome version.

This is configured through the annots.xml file that resides in every genome version directory. Any data set with attribute "load_model" set to "Whole Genome" will automatically load into IGB.

For genomes harvested from UCSC, this foundation gene annotations data typically are from the the UCSC refGene table. If a genome does not have a refGene table, we instead use ensGene or whatever other genes data set looks the most complete and the most useful.

Understand IGB QuickLoad naming conventions

For example, gene models data for the UCSC track named RefSeq Genes is stored in a table called refGene. In IGB QuickLoad, the data file should be named G_species_strain_MMM_YYYY_refGene.bed.gz (a compressed BED detail file) and the title attribute in annots.xml should be RefSeq Genes.

However, the UCSC genome browser sometimes include species names in the title of its mRNA track. IGB QuickLoad data sets should not include species names in the titles. For example, the title of an mRNA track is always "mRNA" and never (for example) "Zebrafish mRNA."

Command-line utilities you'll need need

• tabix from samtools sourceforge site
• bgzip from samtools sourceforge site
• UNIX wget (not installed by default on Mac but available on most other UNIX systems)
• UNIX sort
• UNIX gunzip

Install these in a directory in your PATH.

If you're doing this on a Mac desktop or laptop computer, create a directory called "bin" in your home directory and save all compiled binaries there. Edit your .bash_profile file to include a line like the following to ensure that the shell can find the programs.

export PATH=.:$HOME/bin:$PATH

Step-by-step guide to updating UCSC data sets in IGB QuickLoad

Get the QuickLoad data repo

Check out or update a copy of IGB QuickLoad data and source code directories.

$ svn co https://svn.transvar.org/repos/genomes/trunk/pub/quickload
$ svn co https://svn.transvar.org/repos/genomes/trunk/pub/src quickload_src

If you already have a copy, just update using svn up.

Add quickload_src to your PATH (to run the python code there)

Add quickload_src to your PATH as it contains a python script you'll use to created BED detail files from ordinary BED files. Edit the .bash_profile file as in above:

export PATH=.:$HOME/quickload_src:$HOME/bin:$PATH

Pick a genome to update and change into the genome version directory

Open a Web browser and find the genome in the Table Browser at UCSC

Go to http://genome.ucsc.edu/cgi-bin/hgTables

Update refGene data set

File name: G_species_strain_MMM_YYYY_refGene.gz
Data set title: RefSeq Genes

Get refGene data from UCSC Table Browser

Configure Table Browser

Configure Table Browser with the following settings:

• choose assembly using release date to match it with IGB QuickLoad genome version
• choose group: Genes and Gene Prediction Tracks
• choose track: RefSeq Genes
• choose table: refGene
• choose output format: BED - browser extensible data
• enter output file G_species_MMM_YYYY_refGene.bed.gz
  • The prefix should be THE SAME AS the genome version directory name, with _refGene appended.
• file type returned: choose gzip compressed

Click the "get output" button to download the data

Get gene info and accession info files from NCBI ftp site

$ wget ftp://ftp.ncbi.nlm.nih.gov/gene/DATA/gene2accession.gz
$ wget ftp://ftp.ncbi.nlm.nih.gov/gene/DATA/gene_info.gz

Create BED detail file with gene information

Use ucscToBedDetail.py to create a new BED file with gene symbol and description

For example, do something like this:

$ ucscToBedDetail.py ~/Downloads/G_species_MMM_YYYY_refGene.bed.gz G_species_MMM_YYYY_refGene.bed

This will create a new BED detail file (G_species_MMM_YYYY_refGene.bed) in the current working directory.

Note that for this to work, gene2accession.gz and gene_info.gz must be in your current working directory.

You can also provide these files using options -a and -g if you have saved them to a different location. See the script documentation for details.

Sort, compress, and index

$ sort -k1,1 -k2,2n G_species_MMM_YYYY_refGene.bed | bgzip > G_species_MMM_YYYY_refGene.bed.gz
$ tabix -s 1 -b 2 -e 3 bed G_species_MMM_YYYY_refGene.bed.gz

Validate the file

Check that it has data

$ gunzip -c G_species_MMM_YYYY_refGene.bed.gz | wc -l

The wc command should print the number of lines in the file, which should be equal to the number of rows in the corresponding refGene table. To find out how many rows the refGene table contains, click \"describe table scheme\" in the Table Browser.

Try to open it in IGB

Open file the file and change load mode to whole genome. Click the genome row in the Current Sequence table. You should see something above every chromosome.

If some chromosomes have no data, go back to the table browser and use the region text area to confirm there was no data for the given chromosome.

Check the chromosomes

Make sure the number of chromosomes listedin the file does not exceed the number of chromosomes listed in the genome descriptor file genome.txt:

$ gunzip -c G_species_MMM_YYYY_refGene.bed.gz | cut -f1 | sort | uniq  wc -l
$ wc -l genome.txt

The first line counts the number of unique sequences appearing the first column of the bed file. The second line counts the number of lines in the genome.txt file.

Edit annots.xml

The annots.xml file description for the refGene annotations contains the date the data were downloaded. Edit the file accordingly to reflect today's date.

Check in the new files

Use svn status command to double-check which files you've changed:

$ svn st

It should print something like:

M  G_species_MMM_YYYY_refGene.bed.gz
M  G_species_MMM_YYYY_refGene.bed.gz.tbi
M  annots.xml

M stands for "modified".

Check in the files one-by-one:

svn ci G_species_MMM_YYYY_refGene.bed.gz -m "Enter a message here."
svn ci G_species_MMM_YYYY_refGene.bed.gz.tbi -m "Enter a message here."
svn ci annots.xml -m "Enter a message here."

Update all_mrna

Get mRNA data from UCSC Table Browser

Configure Table Browser

Configure Table Browser with the following settings:

• choose assembly using release date to match it with IGB QuickLoad genome version
• choose group: mRNA and EST tracks
• choose track: SPECIES mRNA
• choose table: all_mrna
• choose output format: selected fields from primary and related tables
• enter output file G_species_MMM_YYYY_all_mrna.psl.gz
  • The prefix should be THE SAME AS the genome version directory name, with _all_mrna appended.
• file type returned: choose gzip compressed

Click "get output" and configure fields

The browser will then show a new screen from which you can select specific fields. For this, select all fields except the field named "bin" (top of the list).

Click "get output" again to download the file.

Create sorted, compressed PSL file minus header

Use grep, sort, and bgzip to make a PSL file:

$ gunzip -c ~/Downloads/G_species_MMM_YYYY_all_mrna.psl.gz | grep -v '^#' | sort -k14,14 -k16,16n | bgzip > G_species_MMM_YYYY_all_mrna.psl.gz

Use tabix to index the sorted, compressed PSL file

$ tabix -s 14 -b 16 -e 17 G_species_MMM_YYYY_all_mrna.psl.gz

Validate the file

Check that it has data

$ gunzip -c G_species_MMM_YYYY_all_mrna.psl.gz | wc -l

The wc command should print the number of lines in the file, which should be equal to the number of rows in the corresponding refGene table. To find out how many rows the refGene table contains, click "describe table scheme" in the Table Browser.

Try to open it in IGB

Open file the file and change load mode to whole genome. Click the genome row in the Current Sequence table. You should see something above every chromosome.

If some chromosomes have no data, go back to the table browser and use the region text area to confirm there was no data for the given chromosome.

Check the chromosomes

Make sure the number of chromosomes listed in the file does not exceed the number of chromosomes listed in the genome descriptor file genome.txt:

$ gunzip -c G_species_MMM_YYYY_all_mrna.psl.gz | cut -f1 | sort | uniq  wc -l
$ wc -l genome.txt

The first line counts the number of unique sequences appearing the first column of the bed file. The second line counts the number of lines in the genome.txt file. They should be the same.

Edit annots.xml

The annots.xml file description for the all_mrna annotations contains the date the data were downloaded. Edit the file accordingly to reflect today's date.

Annots.xml attributes:

All other attributes (e.g., foreground, background, load_hint) should match other UCSC genome release mRNA data sets.

Check in the new files

Use svn status command to double-check which files you've changed:

$ svn st

It should print something like:

M  G_species_MMM_YYYY_all_mrna.psl.gz
M  G_species_MMM_YYYY_all_mrna.psl.gz.tbi
M  annots.xml

M stands for "modified".

Check in the files one-by-one:

svn ci G_species_MMM_YYYY_refGene.bed.gz -m "Enter a message here."
svn ci G_species_MMM_YYYY_refGene.bed.gz.tbi -m "Enter a message here."
svn ci annots.xml -m "Enter a message here."

The message you enter should include the date of the download. This is important because researchers who are using these data will need to know this information in order to publish their results.

Update all_est

Repeat the same steps you used for all_mrna, but instead of choosing Track: SPECIES mRNAs, instead choose SPECIES ESTs, and choose table all_ests.

The data files you create will contain table suffix all_est, e.g., G_species_MMM_YYYY_all_est.psl.gz and its index G_species_MMM_YYYY_all_est.psl.gz.tbi.

The data set file name, title, and description in the annots.xml should be:

file name: G_species_MMM_YYYY_all_est.psl.gz
title: EST
description: UCSC all EST track (date of download)
url: http://www.igbquickload.org/G_species_MMM_YYYY