Arvados Summit Fall 2013 Breakout 1¶

User stories (Jonathan & Ward facilitating)¶

• As an admin, if I change my DB structure, I want Arvados to help me update the config
• As an admin, I want to see the mapping of another dataset to my own
• When I run a job, I want to be able to work as Draft or Final/Real results
• As a consumer of genomic data, I want to visualize my data
• As a commercial leader of a clinical lab, I want to be able to trace quote to cash for diagnostic tests
• I want to be able to know where any file is.
• As a patient or participant, I want to be able to export my data to another study.
• As someone who works with data, I want the genotypic and phenotypic data I use to conform to a standard ontology.
• As a clinician, I want to quantify the uncertainty of the data & analysis underlying my report, so that I and the patient understand the clinical decision more fully.
• As a clinician, I want to run the same experiment on multiple data sets.
• As a lab director and oncologist, I want exome raw reads to called variants to take 15 minutes.
• As a data miner, I want to be able to query all public data without downloading it.
• As a researcher, I want to be able to set up a standard pipeline for a particular type of data set.
• As an informatician, I want all my data to conform to a standard format so that I can analyze across multiple data sets.
• As a clinician, I want to collect & track inbound case data, such as referral letters, ICD-9 diagnosis codes, case summaries, consents, medical reports, and insurance pre-verifications.
• As an informatician, I want to be able to track & manage ICD-9/10 data.
• As a lab director or clinician, I want to share a report with another clinician at another institution.
• As a clinician, if I discover a mutation, I want to share that with an analytical tool or aggregator of data (e.g. GeneInsight).
• As a user, I want to associate ‘keepalive’ metadata to my intermediate data
• As Arvados, I record profiling information that data expiration for intermediate data can be based on
• As an informatician, I can easily manipulate VCF files in parallel (as easy as GNV parallel)
• As a compliance officer, I have structured insight into the consents for my data
• As a researcher, I want to be able to collaborate on big datasets without having to copy them.
• As an informatician, I want to associate metadata with (a section of) my pipelines.
• As a new user, I can browse pipelines for metadata, see how ‘popular’ datasets and pipelines are [‘social features’]

Technical discussion (Tom facilitating)¶

• Test for functionality
• Documentation
  • What can Keep do?
  • High-level functional description
  • How would one replace an existing storage system with Keep?
  • How to migrate?
  • How to MapReduce?
  • Examples
• Databases as input to job
• Permissions
• Audit trail
• Prioritizing jobs - squeaky wheel
• Monitoring - activity & status
• Checkpointing
• Self-starter kit