View Issue Details [ Jump to Notes ]

[ Print ]

Project

Category

View Status

Date Submitted

Last Update

0012772

Visomics

[All Projects] General

public

2012-01-06 15:43

Reporter

Jean-Christophe Fillion-Robin

Assigned To

Priority

normal

Severity

minor

Reproducibility

have not tried

Status

new

Resolution

open

Platform

OS Version

Summary

0012772: Decouple analysis data from user-created Analysis instances

Description

From Brian email - July 19th

- Many of the best planned features require data from multiple Analysis
  * Volcano plot already does this: fold change + t-test
  * Most of Wes's ideas about synergies between analysis and pathways require this
- It would be great to have a common pool of shared analysis outputs / data available
- When a user runs an Analysis, it could fetch from this shared data.
- Using or contributing to the pool is optional. Existing Analysis can continue to compute private data.
- The shared data could be computed lazily, as it's needed by a user-run Analysis.
- Once computed the first time, the data would remain the pool for fast access later.
- Some pieces of shared data, like pathway info, could be generated / prefetched in the background, to make it quickly ready when needed.
- Possible issues could arise if we modify the underlying data.
  * We wanted to be able to hide unimportant Analytes / Experiments. This could invalidate certain existing types of multivariate analysis outputs.
- Memory issues could arise with very large data sets.
- Some types of data may have analysis parameters associated with them. We could store separate data instances for each unique set of parameters (be lazy with this, of course).

Issue History
Date Modified	Username	Field	Change
2012-01-06 15:43	Matthew Bowman	New Issue

Relationships

Notes
There are no notes attached to this issue.