Difference between revisions of "WormBase-Caltech Weekly Calls"
From WormBaseWiki
Jump to navigationJump to searchLine 131: | Line 131: | ||
=== Automated Gene Descriptions === | === Automated Gene Descriptions === | ||
*Ranjana and Valerio working to finish the new pipeline for automated descriptions for WB, will aim to finish them for the next upload, WS269. | *Ranjana and Valerio working to finish the new pipeline for automated descriptions for WB, will aim to finish them for the next upload, WS269. | ||
− | * Working one of the last data types-tissue expression; | + | * Working on one of the last data types--tissue expression; will use the anatomy ontology to perform logical trimming for every cell/anatomy type including neurons (as opposed to using a file for neuronal term groupings taken from Oliver Hobert paper in the old pipeline) |
− | * | + | * Currently playing around with thresholds to see how the sentences look, will feedback any ontology related issues to Raymond |
− | * Working on incorporating feedback from users for information-poor genes (defined as genes with no human orthology and no GO annotations. Will include other types of information suggested by Users such as human ortholog function and protein domains, etc. | + | * Working on incorporating feedback from users for information-poor genes (defined as genes with no human orthology and no GO annotations). Will include other types of information suggested by Users such as human ortholog function and protein domains, etc. |
− | * | + | * Suggestion to exclude the writing and storing of the thousands of automated descriptions to the Postgres database; there is really no advantage in them being in Postgres. |
− | * At the time of generation of the automated descriptions the related .ace files can also be generated; will need to include the 6000+ manual descriptions that live in Postgres. So will need to rethink this part a bit, though skipping Postgres will reduce the number of manual steps in the pipleline and Postgres will have less data that needs to be uploaded and downloaded from future cloud storage. | + | * At the time of generation of the automated descriptions the related .ace files can also be generated; though will need to include the 6000+ manual descriptions that live in Postgres. So will need to rethink this part a bit, though skipping Postgres will reduce the number of manual steps in the pipleline and Postgres will have less data that needs to be uploaded and downloaded from future cloud storage. |
Revision as of 21:06, 23 August 2018
Contents
Previous Years
GoToMeeting link: https://www.gotomeet.me/wormbase1
2018 Meetings
August 2, 2018
AFP
- The AFP pipeline is currently emailing authors from karen's e-mail address
- Use same e-mail account Chris is using for phenotype community curation requests or create a new account for AFP (gmail)
- Can use outreach@wormbase.org for consistency
- May use the PMID in the subject line so e-mails will not be all in the same thread
- Todd and Chris have email credentials
- Chris will send to Valerio, Juancarlos, Daniela, and Kimberly
- Let Valerio and Juancarlos know what pipelines use AFP before they modify
- Do curators still want to receive emails when authors flag their data type?
- We will leave the alert emails as is for now
August 9, 2018
AFP
- Mei Zhen, SAB member suggested that we include disease models in the AFP form.
- The AFP group will work with Ranjana to incorporate it. Ranjana will prepare a mock by next week.
- We will then decide about using the existing afp_humdis tables or creating new ones.
Tazendra
- Shall we move tazendra.caltech.edu to the cloud? Either WormBase cloud or Caltech cloud?
August 16, 2018
Tazendra
- Moving to cloud? To avoid local hardware issues?
- Need to discuss with Juancarlos and Paul S.
- Need to consider logistics; put all of Tazendra functionality on cloud? Keep some things local?
- Postgres in cloud; forms local? Paper pipeline?
- Will consult with Textpresso
ICBO 2018 recap
- POTATO workshop (Phenotype Ontologies Traversing All The Organisms)
- Will work towards generating standardized logical definitions using Dead Simple OWL Design Patterns (DOSDP)
- <Quality> and inheres_in some <Entity> (and has_modifier some <Mod>)
- Exercise: Reconciling logical definitions for apparently equivalent phenotype terms across ontologies (e.g. MP vs. HP)
- Can use Protege to edit the OWL ontology and ROBOT for automating generation of many terms and logical definitions in parallel
- Will try to align WPO to UPheno as best as we can; will depend (at least in part) heavily on alignment with Uberon for anatomy
- Some Uberon alignment challenges: e.g. Fruit fly "tibia" and human "tibia"; human "tibia" parent is "bone" but fly "tibia" is not a bone
- Will participate in Phenotype Ontology Developer's call, every 2 weeks on Tuesdays (9am Pacific, 12pm East coast, 5pm UK)
- Next meeting September 4, 2018
- Crash course in Protege, ROBOT, Ontology Development Kit, using GitHub to help develop OWL ontologies
- PATO needs work
- Questions that arose:
- What should the scope of an ontology term be? Context? Life stage? Conditions? Treatment?
- Being weary of ontology term count explosion; what's the right balance?
- When defining phenotype terms, should the cause be included or only the observation? Maybe causes as a subclass (and assuming the observation includes assessment of cause)
- Some distinction between human phenotype terms and model organism terms: phenotype of individual vs. population
- Will work towards generating standardized logical definitions using Dead Simple OWL Design Patterns (DOSDP)
- Xenbase is trying to develop a phenotype ontology (spoke with Troy Pell, developer)
- Asked about WPO and how we curate
- Lots of plant talks
- Many talks on performing quality checks on ontology development and ontology re-use
- Domain Informational Vocabulary Extraction (DIVE) tool
- Entity recognition/extraction
- Working with two plant journals
- Tries to identify co-occurrence patterns of words
- Web interface and curation tool
- Semantic similarity tools and evaluation of them
WormBase Phenotype Ontology working group
- Chris will send around Doodle poll
- Goal is to discuss creation of logical definitions and alignment of phenotypes for Alliance
August 23, 2018
Alliance tables
- Filtering/sorting priorities
- Open question about which tables on the Alliance website should be prioritized for acquiring sorting and filtering functionality
Worm Phenotype Ontology working group
- Gary S., Karen, Kimberly, and Chris have responded to Doodle poll
- Looks like 12pm Pacific (3pm Eastern) on Thursdays is the time that works for everyone
- May start late on days when WB CIT meeting goes past 12pm Pacific
- May want to start a bit past 12pm to allow west coasters to get lunch, etc.?
- Goals:
- Work on logical definitions for WPO terms
- Consider any restructuring of WPO that would facilitate ontology alignment with other MODs and UPheno
- Could we eventually create a phenotype annotation tool (and term requester) that allows modular expressions of a phenotype observation to lookup existing terms or create new terms with logical definitions based on those modular elements?
Alliance anatomy
- Data quartermasters and expression working group are looking to get updated anatomy-Uberon mappings
- How frequent are data updates at the Alliance? Seems to be every ~2 months
- Anatomy-Uberon mappings will affect phenotype ontology alignments
Automated Gene Descriptions
- Ranjana and Valerio working to finish the new pipeline for automated descriptions for WB, will aim to finish them for the next upload, WS269.
- Working on one of the last data types--tissue expression; will use the anatomy ontology to perform logical trimming for every cell/anatomy type including neurons (as opposed to using a file for neuronal term groupings taken from Oliver Hobert paper in the old pipeline)
- Currently playing around with thresholds to see how the sentences look, will feedback any ontology related issues to Raymond
- Working on incorporating feedback from users for information-poor genes (defined as genes with no human orthology and no GO annotations). Will include other types of information suggested by Users such as human ortholog function and protein domains, etc.
- Suggestion to exclude the writing and storing of the thousands of automated descriptions to the Postgres database; there is really no advantage in them being in Postgres.
- At the time of generation of the automated descriptions the related .ace files can also be generated; though will need to include the 6000+ manual descriptions that live in Postgres. So will need to rethink this part a bit, though skipping Postgres will reduce the number of manual steps in the pipleline and Postgres will have less data that needs to be uploaded and downloaded from future cloud storage.