Opened 19 years ago

Closed 18 years ago

#10 closed task (fixed)

Analysis: Hierarchical clustering

Reported by: Johan Enell Owned by: Jari Häkkinen
Priority: critical Milestone: BASE 2.2
Component: coreplugins Version: trunk
Keywords: Cc:

Description

Hierarchical clustering, bottom-up.

The two closest points are merged and the new cluster is represented by an unweighted(median) or weighted(center of mass) average of the two points in gene expression space. This takes little RAM, allowing you to cluster a large number of genes, but the clustering results could be inferior to those obtained with e.g. average linkage. Also, clustering with absolute correlation as the distance metric does not work very well at all.

Change History (21)

comment:1 by Jari Häkkinen, 19 years ago

Milestone: → BASE 2.0

comment:2 by Nicklas Nordborg, 18 years ago

Priority: major → minor

comment:3 by Jari Häkkinen, 18 years ago

Owner: changed from Johan Enell to Jari Häkkinen
Status: new → assigned

comment:4 by Jari Häkkinen, 18 years ago

Milestone: BASE 2.0 → BASE 2.0+

comment:5 by Jari Häkkinen, 18 years ago

Milestone: BASE 2.x+ → BASE 2.1

comment:6 by Jari Häkkinen, 18 years ago

Priority: minor → blocker

comment:7 by Jari Häkkinen, 18 years ago

Version: → 2.0

comment:8 by Jari Häkkinen, 18 years ago

Milestone: BASE 2.1 → BASE 2.0.1

comment:9 by Jari Häkkinen, 18 years ago

Milestone: BASE 2.0.1 → BASE 2.1

comment:10 by Nicklas Nordborg, 18 years ago

Priority: blocker → critical

comment:11 by Jari Häkkinen, 18 years ago

Milestone: BASE 2.1 → BASE 2.2
Version: 2.0 → trunk

comment:12 by Jari Häkkinen, 18 years ago

Clustering support will be added through TM4s MeV software. Currently bioassay sets must be exported (support exists already in BASE 2.1) for clustering.

comment:13 by Jari Häkkinen, 18 years ago

(In [3096]) Addresses #10, #352, #365. First checkin for JWS use to launch MeV.

comment:14 by Nicklas Nordborg, 18 years ago

(In [3098]) References #10: Hierarchical clustering

Added servlet for downloading bioassay set data in Mev format

comment:15 by Jari Häkkinen, 18 years ago

(In [3100]) Addresses #10 and #352. Now actually using the bioassayset used to trigger MeV.|

comment:16 by Jari Häkkinen, 18 years ago

(In [3101]) Addresses #10 and #352. Fixed problem with non-appearing tool icons in MeV.

comment:17 by Jari Häkkinen, 18 years ago

(In [3102]) Addresses #10 and #352. MeV jsp uses Servlet API to retrieve server information (name,port,...).

comment:18 by Jari Häkkinen, 18 years ago

(In [3105]) Addresses #10 and #352. MeV uses system tmp directory system properties.

comment:19 by Jari Häkkinen, 18 years ago

(In [3106]) Addresses #10 and #352. Removed hard coded indexes for datamatrix retrieval.

comment:20 by Jari Häkkinen, 18 years ago

(In [3110]) Addresses #10 and #352. Unique temporary filenames used at client side.

comment:21 by Jari Häkkinen, 18 years ago

Resolution: → fixed
Status: assigned → closed

Hierarchical clustering is now supported through MeV (​http://www.tm4.org). This is the note regarding MeV:

MeV is now integrated as a Java Web Start application. It requires network and file access so it is signed. You must accept the certificate to be able to run MeV. All algorithms are avaiable for the user but they are currently not fully supported. The main objective was to get Hierarchical clustering working. The integration will be improved in upcoming releases.

Note: See TracTickets for help on using tickets.