Genemania network filtering

[nivedas...@gmail.com]

Hi,

My goal is to filter genemania networks on the basis of edge weights while including all network types (predicted, physical interaction, co-expression etc).

Below is my example dataset:

Gene 1	Gene 2	Weight	Network group	Network
FCGRT	ALB	0.15363139	Physical Interactions	IREF-HPRD
FCGRT	ALB	0.851885	Physical Interactions	IREF-DIP
ALB	GC	0.33333284	Shared protein domains	PFAM
ALB	GC	0.010536404	Co-expression	Dobbin-Giordano-2005
CUBN	ALB	0.70710677	Predicted	I2D-BIND-Rat2Human

Is there a way to:

a) Filter on the basis of 'Weight' column?

b) If so, does higher weight represent high confidence and lower weight represent low confidence?

c) How would I numerically define ranges for edge weights for classification as a) Low confidence b) Medium confidence c) High confidence?  

[Scooter Morris]

Greetings,

a) There is a filter tab in Cytoscape that you can use to filter your network by the Weight column.

b) Higher numbers represent higher confidence.

c) This is a very nuanced question, so I'm going to give you what I would do, but YMMV:  Low: <.5, High: > .8, Medium: >= .5 <= .8  Maybe the GeneMANIA folks might have a different take, but that's generally what I use for STRING networks.

-- scooter

[Gary Bader]

Hi - you can consider higher numbers to be more confident, but they are not confidence values. They represent how useful the networks are to connect your query genes.  So you could try to rank them, but I wouldn't define confidence ranges with those numbers.

Hope that helps,

Gary