Software

Software

Development of tools and applications for Linked Open Data

• Quality assessment of SPARQL endpoints, RDF data and triple stores (cf. yummydata https://github.com/dbcls/bh14/wiki/Yummydata ) [Atsuko(interested)]
  • Automation of utilization, documentation and visualization of RDF data [Atsuko(interested), Kouji(Interested), Yas]
• SPARQL Builder (Kouji, Yasunori, Atsuko) (see also YASGUI, yasgui.org)
  • Improvement of current version (see http://sparqlbuilder.org/) collaborating with other groups (such as Automation of ulitization, Quality assesment, Federated query)
    • Discussion on relationships between SPARQL Builder metadata, ServiceDescription and VoID
    • How to describe relationships of metadata between datasets in the same SPARQL endpoint
  • target endpoints (requests)
    • http://identifiers.org/services/sparql
    • https://www.ebi.ac.uk/rdf/services/biomodels/sparql
    • http://www.genome.jp/sparql/reactionontology/
  • For wrap up: https://github.com/dbcls/bh15/wiki/SPARQLBuilder
• Schema Salad (Peter)
• Federated query via SPARQL endpoints(Hongyan, Atsuko, Jin-Dong, Kouji) (Kieron for Ensembl + PubMed)
• Text search with triple store
  • Embedding ElasticSearch functions in RDF store to enhance query function and performance, such as autocompletion and so on (G. Fu)
• the BioVirtuoso Docker data containers (HiroMishima, https://github.com/misshie/bio-virtuoso )[T Nakazato(interested)]
  • Bio2RDF - MichelD
  • OrphaData, better RDFized HPO Annotation
• Common Workflow Language, portable workflows, container standards (Peter, Colin, Tazro)
  • CWL tutorial
  • wrapping tools and writing workflows using CWL
  • containers, runtime configuration
  • writing CWL implementations (JS, Ruby, Java?)
  • tool & workflow registries, workflow metadata ontologies (Dublin core, EDAM, DOAP)
  • visualisation
  • software discovery
  • RDF/SPARQL explorer and visualizer (Naoki)
• SMART API - semantic annotation of web APIs http://smartapi.info - (MichelD) [Nick interested]
  • develop documentation
  • annotate biohack15 APIs
• Semantic Wetlab (Erick Antezana, Alexander Garcia, Tazro Ohta, Jean-Luc Perret)
  • Ontologies for representing investigations [OBI?][SIO?][EDAM][ISA]
    • experimental design
    • rdf specification for workflows
    • tools for designing, planning and running experiments (how good should look like, use case)
  • Report BH 2015
• Reproducible Software and data deployment (Pjotr Prins)
  • GUIX for databases (Jerven Bolleman, Raoul)
  • Software discovery (Pjotr, Raoul)
  • Ruby biogem support (Pjotr, Raoul, Naohisa)
  • GUIX for UniProt RDF releases (Jerven)
  • GUIX UniProt virtuoso local builds
    • With minimal auto tuning for memory
  • Reproducible software and data with Dat, hyperos.io (Bruno, Tazro)
    • Bionode pipeline inside, visualization with BioJS, nyaplot, D3
• Visualisation
  • D3 visualisation work group (Toshiaki, Naoki, Pjotr, Peter, Bruno, MichelD)
• Open-Bio
  • BioRuby (Naohisa Goto,)
    • Sequence metadata RDF conversion (consider using RDF.rb --Arto Bendiken)
      • Revisit Bio::Sequence internal data structure for better RDFize
      • https://github.com/dbcls/rdfsummit/tree/master/insdc2ttl (Toshiaki Katayama, Takatomo Fujisawa)
    • UniProt RDF module (alternative to SPTRParser)
      • A JSON-LD based prototype, https://github.com/nakao/uniprot-rdf-module (Mitsu)
    • Biogems
      • biogem simplification
        • Switch from jeweler to bundler
    • bio-keggapi (for REST API, SOAP is deprecated now)
    • Split independent functional components from bioruby core to biogems
      • Components having external dependencies
        • bioruby-phyloxml ?
          • Based on https://github.com/csw/bioruby-phyloxml
          • Adding support for Nokogiri (XML parser) based on https://github.com/csw/bioruby/tree/phyloxml-nokogiri
        • bioruby-biosql ?
• Crick-chan - a question answering system (Kazuharu Arakawa, Kotone Itaya)
  • Crick-chan

Day 1

Bio-virtuoso

Participants: Hiro Mishima, Jeremy Nguyen Xuan, Tudor Grosa

See BioVirtuoso

Day 2

project

Participants: ...

• CWL
  • Tutorial given by Peter Amstutz. Attended by Benedict Paten, Bruno Viera, Alex Garcia, Tazro Ohta, others++
  • Discussion of how to combine CWL, Docker Hub, Elixir Tool Registry to provide a central repository for bioinformatics tools that can be directly downloaded and executed in workflows (no installation needed)
  • Annotating CWL files with metadata
  • Ways of running Docker when the IT staff doesn't want to run Docker (solution: run Docker inside VM)
  • Rebasing CWL draft 3 on Salad schema to support linked data annotations

SPARQL Builder

Participants: Atsuko, Kouji, Yasunori ...

• Re-design for SPARQL Builder Matadata(SBM)
• Setup SPARQL endpoint for SBM, etc.

Day 3

SPARQL Builder

Participants: Atsuko, Kouji, Yasunori ...

• new version of specification of SPARQL Builder Matadata(SBM) was released http://www.sparqlbuilder.org/doc/sbm_2015sep/
• Thank to Arto, Dydra can automatically generate the metadata.
• crawling for LSDB archive rdf
• Since problems are found, we asked the author of the crawler to fix them.

CWL

• Docker with Guix package manager (Bruno, Pjotr, Raoul)
  • https://hub.docker.com/r/bmpvieira/guix/
  • https://github.com/bmpvieira/Dockerfiles/blob/master/guix/Dockerfile
• Started installing CWL with Guix (Bruno, Pjotr)
• CWL Streaming interface (Peter, Bruno)
• Started playing with Bionode JavaScript and CWL streaming (Bruno)

RDF::VCF

• https://github.com/ruby-rdf/rdf-vcf

Day 4

• Federated query service: http://fsearch.dbcls.jp/bio2RDF/

SPARQL Builder

Participants: Atsuko, Kouji, Yasunori ...

• Because anyone who are users of Dydra can generate SB metadata for their DB, we started to develop an interface for uploading SB metadata.
• alpha version of SPARQL Builder for LSDB archive
  • http://www.sparqlbuilder.org/dba/

Day 5

BioRuby

• Bio::PhyloXML is moved to https://github.com/bioruby/bioruby-phyloxml (Naohisa)
  • Bio::PhyloXML is now removed from BioRuby core (https://github.com/bioruby/bioruby)
  • bio-phyloxml gem will soon be released (stopping because of gem upload privilege)
• BioSQL support will be moved to https://github.com/bioruby/bioruby-biosql
• Discussing to improve BioRuby's Bio::Seqnece and https://github.com/dbcls/rdfsummit/tree/master/insdc2ttl by using RDF.rb

Day 6

Embedding Elasticsearch into SPARQL and Cypher

Participants: Fu Gang, Jeremy

Inspired by Aber-owl: embedded DL query into SPARQL (http://aber-owl.net/aber-owl/sparql/) Elasticsearch allows synonym search (doc: https://www.elastic.co/guide/en/elasticsearch/guide/current/using-synonyms.html), spell correction (doc: https://www.elastic.co/guide/en/elasticsearch/reference/current/search-suggesters-term.html), and phrase match (doc: https://www.elastic.co/guide/en/elasticsearch/guide/current/phrase-matching.html).

Experiment setup:

1. Elasticsearch index of PubMed title (data file 2.4G, index file 1.7G)

2. search phrase: "drug treat disease" returned 415 230 records; "gene mutation cause disease" returned 598 057 records.

3. embed the results into sparql query using 'values' keyword: challenges cannot allow too many records.

Improve performance of a query

Typical query performed in our group:

START disease = node:node_auto_index(iri={disease_id}) MATCH path = (disease)<-[:subClassOf*0..]-(diseaseSubclass)-[:hasPhenotype]->(phenotype) RETURN distinct phenotype

Expensive to run in big/deep ontologies.

Idea: Index in ElasticSearch the subclasses and superclasses of all the nodes, and delegate the expensive part of the query to ElasticSearch instead of performing it all in Cypher.

Experiment setup:

• Subset of ontologies used in the Monarch Initiative (2.1G, 74k nodes)
• Query runtime was divided by half.

VCF to RDF Mapping

Reproducible and distributable software and data

• Use Bionode for streamable workflow (Bruno)
  • Reason: Node.js events and Streams are very flexible for scalable pipelines and workflows
• Use Guix for package management
  • Reason: Reproducible software installation, dependency management, toolchain independent from OS.
  • Docker container with Guix (Bruno, Pjotr, Raoul)
• Use CWL for tools integration
  • Reason: Integration between several bioinformatics tools, JSON stdin and stdout (easy to integrate with Bionode).
  • Streamable CWL (Peter, Bruno)
  • CWL Guix package (Pjotr, Bruno)
• Run Docker tarball with Dat/Hyperos (Bruno)
  • Reason: Run tarball in non-Docker environments (e.g., HPC)
• Discussion around standards and distribution of containers for bioinformatics (Benedict, Peter, Bruno)
• Implemented CWL support within Toil (pip install toil), a scaleable workflow execution engine (Peter, Benedict)