Software Life Cycle: 1. Updating The Development Server

THIS DOCUMENT IS NOW DEPRECATED. PLEASE REFER TO THE PROJECT DOCUMENTATION MAINTAINED ON GOOGLE DRIVE.

https://docs.google.com/a/wormbase.org/document/d/1oPpj8d5gibUc-gpUZorl6ETT5baE6mp-v2bMedKauiA/edit#

Overview

This document describes the process of staging a new release of WormBase on the development server.

The automated staging pipeline consists of:

• a harness that handles logging, error trapping, and basic shared functions
• a suite of modules -- one per step -- that implement the step or make calls to helper scripts
• helper scripts in Perl or shell that assist in implementation

Control of the pipeline: You can use the pipeline in several ways:

• Launch the full pipeline via the control script, the preferred and automated method.
• Run individual steps in the context of the pipeline using control scripts in steps/, useful if the pipeline fails at a specific point.
• Directly run helper scripts outside of the logging facilities of the pipeline, useful if you need to rebuild something quickly.

Document Conventions

The current development server is

 wb-dev: wb-dev.oicr.on.ca (FQDN); aka: dev.wormbase.org

When indicated, substitute WSXXX for ${RELEASE}.

System paths referred to in this document:

      FTP : /usr/local/ftp/pub/wormbase
 WORMBASE : /usr/local/wormbase
    ACEDB : /usr/local/wormbase/acedb

Staging Pipeline Code

The update pipeline code is available in the website-admin module on github:

tharris> git clone git@github.com:WormBase/website-admin.git
tharris> cd website-admin/update

 lib/                  -- the shared library suite that handles updates.
 staging/          -- code related to staging data on the development site. 
 production/     -- code related to the releasing data/code into production.

The contents are:

  logs/            -- the logs directory for each step/update
  bin/              -- Perl scripts for manually launching individual steps.
  README.txt -- directory listing
  updatelog.conf  -- a configuration file for the update process
  update.sh  -- master script that fires off each step of the pipeline
  util/       -- various helper scripts for the update process

Running the Update Pipeline

Log Files

The Staging Pipeline creates informative logs for each step of the process. Logs are located at:

 /usr/local/wormbase/logs/staging/WSXXX
     master.log  -- Master log tracks all steps; useful for a meta-view of the pipeline. Contains INFO, WARN, ERROR, and FATAL messages.
     master.err  -- Master error log tracks ERROR and FATAL messages encountered across all steps.

Each individual step creates its own log file capturing STDERR and STDOUT containing informative messages from the pipeline. These are useful for tracking progress and triaging problems. For example:

 /usr/local/wormbase/logs/staging/WSXXX/build_blast_databases/
      step.log   -- step-specific log tracking everything from TRACE on up.
      step.err    -- step-specific error log tracking ERROR and FATAL messages. Good place to check if a step breaks.

Executing the Pipeline

A single script fires off all steps of the process. You should run it inside a screen.

 tharris> screen
 tharris> ./stage_via_pipeline.pl WSXXX
   (to disconnect your screen)
 tharris> ^a ^d
   (to resume your screen)
 tharris> screen -r   
 

Monitor progress of the update by following the master log file:

tharris> tail -f /usr/local/wormbase/logs/staging/WSXXX/master.log

 screen command reference

Update Steps

The steps that comprise the pipeline, the script to launch them, and the module that implements are listed below.

step	control script	module
Mirror a new release	steps/mirror_new_release.pl (manual)	W::U::Staging::MirrorNewRelease
Unpack ACeDB	steps/unpack_acedb.pl (manual)	W::U::Staging::UnpackAcedb
Create BLAST databases	steps/create_blast_databases.pl	W::U::Staging::CreateBlastDatabases
Create BLAT databases	steps/create_blat_databases.pl	W::U::Staging::CreateBlatDatabases
Load Genomic GFF databases	steps/load_genomic_gff_databases.pl	W::U::Staging::LoadGenomicGFFDatabases
Unpack and Load the ClustalW database	steps/unpack_clustalw_database.pl	W::U::Staging::UnpackClustalWDatabase
Compile Gene Summary resources	steps/compile_gene_resources.pl	W::U::Staging::CompileGeneResources
Compile Ontology resources	steps/compile_ontology_resources.pl	W::U::Staging::CompileOntologyResources
Compile Orthology resources	steps/compile_orthology_resources.pl	W::U::Staging::CompileOrthologyResources
Create commonly requested datasets	steps/dump_annotations.pl	W::U::Staging::DumpAnnotations
Go Live	steps/go_live.pl	W::U::Staging::GoLive
Convert GFF2 To GFF3	steps/convert_gff2togff3.pl	W::U::Staging::ConvertGFF2ToGFF3
Precache content	steps/precache_content.pl	W::U::Staging::PrecacheContent

• Compile orthology resources
• Compile interaction resources
• Build and load GFF patches
• Create a GBrowse-driven genetic map
• Create a GBrowse-drive physical map

Purge old releases

Clear out disk space by throwing away old releases.

./steps/purge_old_releases.sh WSXXX   // release to purge; clears out acedb, mysql, support DBs, and staging FTP

Mirror a new release

New releases are mirrored directly from the Hinxton FTP site to the primary WormBase FTP site hosted on wb-dev:/usr/local/ftp. This process is run via cron but can also be run manually.

 # Mirror the next incremental release newer than what we already have:
 # Cron: 
 ./steps/mirror_new_release.pl

 # Or mirror a specific release: 
 ./steps/mirror_new_release.pl --release WS150   // Mirror the WS150 release to /usr/local/ftp/pub/wormbase/releases/WS150

Unpack Acedb

Unpack AceDB from the new release. Customize the new installation with skeletal files located at /usr/local/wormbase/website/classic/wspec. You will need approximately 25 GB of disk space per release.

via pipeline: ./steps/unpack_acedb.pl ${RELEASE}
via helper : helpers/unpack_acedb.sh ${RELEASE}
  Input : Files staged at ${FTP}/releases/${RELEASE}/species
 Output : Unpacked AceDB files in ${ACEDB}/wormbase_${RELEASE} 

When complete, you should have a new acedb directory containing:

   -- database
   -- wgf
   -- wquery
   -- wspec

Test the database by:

> ps -ax | grep acedb ## to get acedb process number
> kill -9 {AceDB proc number} ## stop current acedb process
> sudo /etc/init.d/xinetd restart 
> saceclient localhost -port 2005

Create BLAST databases

Build nucleotide and protein BLAST databases for species with genomic sequence and conceptual translations. In addition, for C. elegans and C. briggsae, we build blast databases for ESTs and "genes" (actually clones).

  Usage : ./steps/create_blast_databases.pl ${RELEASE}
  Input : Genomic sequence and protein FASTA files staged at:
             ${FTP}/releases/species/${SPECIES}.${RELEASE}.genomic.fa.gz
             ${FTP}/releases/species/${SPECIES}.${RELEASE}.protein.fa.gz
             Gene and EST sequences derived from AceDB
 Output : BLAST databases in ${WORMBASE}/databases/${RELEASE}/blast/${SPECIES}.

Create BLAT databases

Build BLAT databases of genomic sequence.

  Usage : ./steps/create_blat_databases.pl ${RELEASE}
  Input : Genomic sequence FASTA files staged at
             ${FTP}/releases/species/${SPECIES}/${SPECIES}.${RELEASE}.genomic.fa.gz
 Output : BLAT .nib files in ${WORMBASE}/databases/${RELEASE}/blat/${SPECIES}

Load genomic GFF annotations

Convert GFF files into Bio::DB::GFF (GFF2) or Bio::DB::SeqFeature::Store (GFF3) databases.

 Usage : ./steps/load_genomic_gff_databases.pl ${RELEASE}
 Input : GFF and FASTA files staged at:
           GFF : ${FTP}/releases/species/${SPECIES}/${SPECIES}.${RELEASE}.gff[2|3].gz
           DNA : ${FTP}/releases/species/${SPECIES}/${SPECIES}.${RELEASE}.genomic.fa.gz

Unpack and Load the CLUSTALW database

Usage: ./steps/unpack_clustal_database.pl {WSRELEASE}
Input: ${FTP}/releases/${RELEASE}/COMPARATIVE_ANALYSIS/wormpep${RELEASE}.clw.sql.bz2
Output: a new mysql database called clustal_${RELEASE}

Pre-compiled file documentation for Beta

NOTE: no need to pre-compile Gene and Interaction data resources any more

Compile Ontology Resources

TODO: This step is just copying over files, could be avoided but maybe file access issue?

 In Wormbase.conf 
 association_count_file = /usr/local/wormbase/databases/%s/ontology/%s_association.%s.wb
  

copy over the mirrored ontology files under ftp directory, these files are used for calculating the counts of associated terms in Ontology Browser.

  Usage : ./steps/compile_ontology_resources.pl ${RELEASE}
  Input : WB files staged at: /usr/local/ftp/pub/wormbase/releases/WSXXX/ONTOLOGY

• anatomy_association.RELEASE.wb
• gene_association.RELEASE.wb
• phenotype_association.RELEASE.wb

 Output : to ${WORMBASE}/database/${RELEASE}/ontology:

• anatomy_association.RELEASE.wb
• gene_association.RELEASE.wb
• phenotype_association.RELEASE.wb

Compile Orthology Resources

  Usage : ./steps/compile_orthology_resources.pl ${RELEASE}
  Input :  omim.txt and morbidmap files from OMIM

  Intermediate: gene2omim.txt  (query AceDB and build the worm gene to human disease relationship based on its human ortholog)

 Output : Files ${WORMBASE}/databases/${RELEASE}/orthology 
 Disease.ace (load into Xapian, use for disease/search and disease page)

Compile Gene Resources

BROKEN

Create precompiled gene page files specifically to populate the Phenotype tables.

  Usage : ./steps/compile_gene_resource.pl ${RELEASE}
  Input : AceDB data
 Output : Files ${WORMBASE}/databases/${RELEASE}/gene

• gene_rnai_pheno.txt (gene/gene)
• gene_xgene_pheno.txt (gene/gene)
• phenotype_id2name.txt (gene/gene)
• rnai_data.txt (gene/gene)
• variation_data.txt (gene/gene)

Compile Ontology Resources

TODO: This step relies on a number of external helper scripts that should ALL be folded into CompileGeneResources. They are located at

staging/helpers/gene_summary 

Take the mirrored ontology files and compile them into the databases for the ontology searches.

  Usage : ./steps/compile_ontology_resources.pl ${RELEASE}
  Input : OBO files staged at: /usr/local/ftp/pub/wormbase/releases/WSXXX/ONTOLOGY
          compiled data files from Compile Gene Resources step

• anatomy_association.RELEASE.wb
• anatomy_ontology.RELEASE.obo
• gene_association.RELEASE.wb
• gene_association.RELEASE.wb.cb
• gene_association.RELEASE.wb.ce
• gene_association.RELEASE.wb.cjp
• gene_association.RELEASE.wb.ppa
• gene_association.RELEASE.wb.rem
• gene_ontology.RELEASE.obo
• phenotype_association.RELEASE.wb
• phenotype_ontology.RELEASE.obo

 Output : to ${WORMBASE}/database/${RELEASE}/ontology:

• id2association_counts.txt (ontology/tree_lister)
• id2name.txt (ontology/tree_lister)
• id2parents.txt (ontology/tree_lister)
• id2total_associations.txt (ontology/tree_lister)
• name2id.txt
• search_data.txt
• parent2ids.txt (ontology/tree_lister)

Compile Orthology Resources

Create precompiled orthology and disease display and search related files

This MUST be run after the ontology step above.

  Usage : ./steps/compile_orthology_resources.pl ${RELEASE}
  Input : AceDB data, omim.txt and morbidmap files from OMIM, ontology resources files

• gene_association.$RELEASE.wb.ce

  Intermediate: 

• all_proteins.txt
• disease_page_data.txt
• full_disease_data.txt
• gene_id2go_bp.txt
• gene_id2go_mf.txt
• gene_id2phenotype.txt
• gene_list.txt
• hs_proteins.txt
• last_processed_gene.txt
• omim2disease.txt
• omim_id2go_ids.txt
• omim_id2phenotypes.txt
• omim_id2disease_synonyms.txt
• omim_reconfigured.txt
• ortholog_other_data.txt
• ortholog_other_data_hs_only.txt

 Output : Files ${WORMBASE}/databases/${RELEASE}/orthology (summary page using files in parenthesis)

• disease_search_data.txt (orthology/search)
• gene_id2omim_ids.txt (orthology/disease)
• go_id2omim_ids.txt (orthology/disease,ontology/gene)
• go_ids2descendants.txt (orthology/gene)
• hs_ensembl_id2omim.txt (orthology/gene)
• id2name.txt (orthology/disease, orthology/gene)
• name2id.txt (orthology/disease)
• omim_id2all_ortholog_data.txt (orthology/disease)
• omim_id2disease_desc.txt (orthology/disease)
• omim_id2disease_name.txt (orthology/disease,ontology/gene)
• omim_id2disease_notes.txt (orthology/disease)

• omim_id2disease.txt (orthology/gene)
• omim_id2gene_name.txt (orthology/search)

Compile Interaction Data

DEPRECATED. NO NEED TO MIGRATE THIS INTO THE NEW STAGING PIPELINE.

Create precompiled gene page files specifically to populate interaction listing pages.

  Usage : ./steps/compile_interaction_data.pl ${RELEASE}
  Input : AceDB interaction data
 Output : Files ${WORMBASE}/databases/${RELEASE}/interaction

• compiled_interaction_data.txt

Convert GFF2 into GFF3

 Usage: ./steps/convert_gff2_to_gff3.pl ${RELEASE}

Create files of commonly requested datasets

 Usage: ./steps/dump_annotations.pl {WSRELEASE}
Output: datasets in ${FTP}/releases/${RELEASE}/annotations and species/annotations

The staging harness will automatically run scripts in annotation_dumpers/*. These scripts should abide by the following conventions:

   1. Be located in update/staging/annotation_dumpers                                                                                                      
   2. Be named either                                                                                                                                      
          dump_species_*   for species level data (like brief IDs)                                                                                          
          dump_resource_*  for resource level data (like laboratories)                                                                                      
   3. Follow existing examples, including available parameters.                                                                                            
   4. Dump to STDERR and STDOUT.                                                                                                                           
   Notes:                                                                                                                                                  
                                                                                                                                                            
   1. dump_species_* will be called for each species managed by WormBase                                                                                   
      and will end up in                                                                                                                                   
         ${FTP_ROOT}/releases/[RELEASE]/species/[G_SPECIES]/annotation/[G_SPECIES].[RELEASE].[DESCRIPTION].txt                                            
      dump_resource_* will be called once and end up in                                                                                                    
         ${FTP_ROOT}/datasets-wormbase/wormbase.[RELEASE].[DESCRIPTION].txt                                                                   
   2. The filename will be created by stripping off dump_species_ or dump_resource_.                                                                       
       Species specific resources will be prepended with the appropriate species.

Create a GBrowse-driven genetic map

Notes...

 Usage: ./steps/load_gmap_gffdb.pl ${RELEASE}

Go Live

steps/go_live.pl WSXXX

This script will

• create a series of symlinks in the FTP site (for example, to maintain the virtually organized species/ directory)
• create "current" symlinks in the FTP site for easy access.
• adjust symlinks to mysql GFF databases updated this release.
• adjust the symlink at /usr/local/wormbase/acedb/wormbase -> to point to the new wormbase_WSXXX unpacked acedb.
• Sync the staging FTP site to the production FTP site.

If you omit the WSXXX on the command line, the script will simply organize the virtual directories on the ftp site up to and including the current release. MySQL and AceDB symlinks will not be created.

Branch the web code

For each major WS release, create a corresponding branch in the git repository. We branch the moment a new release is staged so that we can begin development for that model. This can be done from any repository.

staging> cd /usr/local/wormbase/website/production
staging> git pull
// Creating a tag...
// staging> git tag -a -m "WSXXX" WSXXX HEAD
// staging> git push --tags
// Create a new branch, one tracking the remote master repository
staging> git branch --track WSXXX origin/master
staging> git branch // list all branches
// Push the branch to the remote repository
staging> git push origin WSXXX
staging> git push

Steps to execute after a release has been staged

Precache content

Once a release has been successfully staged and tested, we pre-cache select computationally intensive content to a CouchDB instance located on the development server.

Precaching works as follows.

1. The primary Catalyst configuration file is read.

2. For each widget set to "precache = true" in config, REST requests will be constructed against staging.wormbase.org. This will be running the NEW version of WormBase.

3A. The webapp returns HTML; the precache script stores it in the reference (production) couchdb.

   OR

3B. The web app on staging.wormbase.org will automatically cache result in the reference couchdb (currently web6); the couchdb that is written to can be configured in wormbase.conf.

4. The reference couchDB will then be replicated during production release to each node, scaling horizontally.

5. During a production cycle, additional content will be stored in the reference couchdb; this is synced periodically to each node.

See CouchDB for details.

Purge old releases

To purge previous releases from the production and staging nodes,

staging/steps/purge_old_releases.pl --release WSXXX

This will remove the following:

/usr/local/wormbase/acedb/wormbase_WSXXX
/usr/local/wormbase/databases/WSXXX
/usr/local/mysql/data/WSXXX

And on the staging host

/usr/local/ftp/pub/wormbase/releases/WSXXX

Compiled file documentation and plans

Step	File	Description	WB2 update
Compile Gene Resources	gene_rnai_pheno.txt	many-to-many listing of Gene_ids to RNAi_ids and Related phenotype ID. (or not) Used in the classic gene summary for phenotype tables	TODO: update the appropriate method in Gene.pm to pull data for given gene directly from Ace
Compile Gene Resources	gene_xgene_pheno.txt	many-to-many listing of Gene_ids to Transgene_ids and Related phenotype ID. (or not) Used in the classic gene summary for phenotype tables	TODO: update the appropriate method in Gene.pm to pull data for given gene directly from Ace
Compile Gene Resources	phenotype_id2name.txt	listing of Phenotype_ids to Phenotype names. Used in the classic gene summary for phenotype tables in order to obviate the extraction of individual phenotype objects and their names	TODO: function will be deprecated since individual phenotype objects will be extracted
Compile Gene Resources	rnai_data.txt	listing of RNAi data for the RNAi table in gene/gene Used in the classic gene summary for RNAi tables	TODO: update the appropriate method in Gene.pm to pull data for given gene directly from Ace
Compile Gene Resources	variation_data.txt	many-to-many listing of Gene_ids to RNAi_ids and Related phenotype ID. (or not) Used in the classic gene summary for phenotype tables	TODO: update the appropriate method in Gene.pm to pull data for given gene directly from Ace
Compile Ontology Resources	id2association_counts.txt	listing of ontology object ids (GO, Anatomy_term, Phenotype) to the number of annotations to the term used in tree_lister (browser)	retain for browser, move into tied hash?
Compile Ontology Resources	id2name.txt	listing of ontology object ids (GO, Anatomy_term, Phenotype) to the term used in tree_lister (browser)	retain for browser, move into tied hash?
Compile Ontology Resources	id2parents.txt	one-to-many listing of ontology object ids (GO, Anatomy_term, Phenotype) to the parent terms and respective relationship used in tree_lister (browser)	retain for browser, move into tied hash?
Compile Ontology Resources	id2total_associations.txt	listing of ontology object terms (GO, Anatomy_term, Phenotype) to the id used in tree_lister (browser)	retain for browser, move into tied hash?
Compile Ontology Resources	search_data.txt	pipe-delieneated data on each term including synonyms and annotations Used in GO, Anatomy_term, and Phenotype searches	To be superceded by Xapian search
Compile Ontology Resources	parent2ids.txt	one-to-many listing of ontology object ids (GO, Anatomy_term, Phenotype) to their immediate descendants term ids used in tree_lister (browser)	retain for browser, move into tied hash?
Compile Orthology Resources	disease_search_data.txt	Pipe delineated file containing details on the diseases extracted from OMIM used in disease search	Use data for Xapian search; work with Abby
Compile Orthology Resources	gene_id2omim_ids.txt	one-to-many listing of gene_ids to omim IDs used in orthology/disease	Keep for disease object
Compile Orthology Resources	go_id2omim_ids.txt	one-to-many listing of gene_ids to omim IDs used in orthology/disease and ontology/gene	useful for further paralog data expansion and integration
Compile Orthology Resources	go_ids2descendants.txt	one-to-many listing of go ids to its list of the go ids of its descendants plan was to use this data for paralog display in orthology/gene	useful for further paralog data expansion and integration
Compile Orthology Resources	hs_ensembl_id2omim.txt	one-to-one listing of hs ensembl ids to omim ids used in orthology/gene	disease UI
Compile Orthology Resources	id2name.txt	listing of ontology object ids (GO, Anatomy_term, Phenotype) to the term used in orthology/disease & orthology/gene	useful for further paralog data expansion and integration(?)
Compile Orthology Resources	name2id.txt	listing of ontology object terms(GO, Anatomy_term, Phenotype) to the id used in orthology/disease	useful for further paralog data expansion and integration(?)
Compile Orthology Resources	omim_id2all_ortholog_data.txt	pipe delineated file containing details of the ortholog associated with the omim id used in orthology/disease	use to generate Xapian data; work with Abby
Compile Orthology Resources	omim_id2disease_desc.txt	one-to-one listing of omim ids and the disease description used in orthology/disease	use in Disease object model and UI
Compile Orthology Resources	omim_id2disease_name.txt	one-to-one listing of omim ids and the disease name used in orthology/disease	use in Disease object model and UI
Compile Orthology Resources	omim_id2disease_notes.txt	one-to-one listing of omim ids and the disease notes from omim used in orthology/disease	use in Disease object model and UI
Compile Orthology Resources	omim_id2disease.txt	one-to-one listing of omim ids and the disease names used in orthology/disease	use in Disease object model and UI
Compile Orthology Resources	omim_id2gene_name.txt	one-to-many listing of omim ids to gene names used in orthology/search	probably deprecate in updating Disease object model