search

BaderLab / EnrichmentMapApp

The EnrichmentMap Cytoscape App allows you to visualize the results of gene-set enrichment as a network.
http://apps.cytoscape.org/apps/enrichmentmap
GNU Lesser General Public License v2.1
31 stars 12 forks source link

EM version 3.4.0-SNAPSHOT" Default style node color: -log10(pvalue)_max (enhancement) #538

Closed veroniquevoisin closed 4 months ago

veroniquevoisin commented 4 months ago

Thanks for adding this default node color style . For GSEA or generic format with NES +1 and -1, is it possible to add a column attribute that would look like: -log10(pvalue)_max * sign(NES) so that we get blue and red nodes? Screenshot 2024-02-26 at 2 40 02 PM

veroniquevoisin commented 4 months ago

Here is how I changed the style:

Screenshot 2024-02-28 at 9 43 00 AM
mikekucera commented 4 months ago

Hi Veronique, I have a question...

The mapping for -log10(pvalue) in your screenshot uses the range (2.16, 2.92) which does not include zero and has size 2.92 - 2.16 = 0.76

If we make the range go from (-2.92, 2.92) then the range will include zero and have a size more than double the original size. This will affect the color interpolation because the gradient covers a much larger range.

Is this ok? And should the range for -log10(pvalue) go from zero to the max value?

veroniquevoisin commented 4 months ago

Hi Mike,

This one would be ideal: 0 in the middle:If we make the range go from (-2.92, 2.92) then the range will include zero and have a size more than double the original size. This will affect the color interpolation because the gradient covers a much larger range.

Veronique

On Mon, Mar 4, 2024 at 6:30 PM Mike Kucera @.***> wrote:

Hi Veronique, I have a question...

The mapping for -log10(pvalue) in your screenshot uses the range (2.16, 2.92) which does not include zero and has size 2.92 - 2.16 = 0.76

If we make the range go from (-2.92, 2.92) then the range will include zero and have a size more than double the original size. This will affect the color interpolation because the gradient covers a much larger range.

Is this ok? And should the range for -log10(pvalue) go from zero to the max value?

— Reply to this email directly, view it on GitHub https://github.com/BaderLab/EnrichmentMapApp/issues/538#issuecomment-1977649285, or unsubscribe https://github.com/notifications/unsubscribe-auth/ABZYHS6TVLTUAWPH74ASFUTYWT72JAVCNFSM6AAAAABD22GK4CVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMYTSNZXGY2DSMRYGU . You are receiving this because you authored the thread.Message ID: @.***>

mikekucera commented 4 months ago

For the regular -log10(pvalue) chart should it go from 0 to 2.92 in this case? I think that would be consistent. Switching to the -log10(pvalue)*sign(NES) chart wouldn't affect the color of nodes that have a positive NES, which makes sense to me.

veroniquevoisin commented 4 months ago

Yes, what you are saying makes sense to me. Veronique

On Mon, Mar 4, 2024 at 7:33 PM Mike Kucera @.***> wrote:

For the regular -log10(pvalue) chart should it go from 0 to 2.92 in this case? I think that would be consistent. Switching to the -log10(pvalue)*sign(NES) chart wouldn't affect the color of nodes that have a positive NES, which makes sense to me.

— Reply to this email directly, view it on GitHub https://github.com/BaderLab/EnrichmentMapApp/issues/538#issuecomment-1977729918, or unsubscribe https://github.com/notifications/unsubscribe-auth/ABZYHS6KNHZY2I7BZC5TGWDYWUHEBAVCNFSM6AAAAABD22GK4CVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMYTSNZXG4ZDSOJRHA . You are receiving this because you authored the thread.Message ID: @.***>

mikekucera commented 4 months ago

Hi Veronique, another question.

The mapping was to the -log10(pvalue)_max column, which contains the maximum -log10(pvalue) when there's more than one data set.

With NES I think its a bit harder to deal with. Its possible the NES could be positive in one data set and negative in another. Its not clear to me what to do in this situation.

Options... 1) Multiply by the sign of the largest NES value 2) Only multiply by sign(NES) when the network has exactly one data set, otherwise just use -log10(pvalue)_max for networks that have more than one data set. 3) others options?

veroniquevoisin commented 4 months ago

Hi Mike, from your suggestions I would go with option2. Veronique

On Tue, Mar 5, 2024 at 10:18 AM Mike Kucera @.***> wrote:

Hi Veronique, another question.

The mapping was to the -log10(pvalue)_max column, which contains the maximum -log10(pvalue) when there's more than one data set.

With NES I think its a bit harder to deal with. Its possible the NES could be positive in one data set and negative in another. Its not clear to me what to do in this situation.

Options...

  1. Multiply by the sign of the largest NES value
  2. Only multiply by sign(NES) when the network has exactly one data set, otherwise just use -log10(pvalue)_max for networks that have more than one data set.
  3. others options?

— Reply to this email directly, view it on GitHub https://github.com/BaderLab/EnrichmentMapApp/issues/538#issuecomment-1979012073, or unsubscribe https://github.com/notifications/unsubscribe-auth/ABZYHS2Z72KNHXOZZZTUHTDYWXO35AVCNFSM6AAAAABD22GK4CVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMYTSNZZGAYTEMBXGM . You are receiving this because you authored the thread.Message ID: @.***>

mikekucera commented 4 months ago

7e87f18, 512dc4c00b21019153f00b01f899fbe541fd32f4, 00443a02ab69b0e40ea73627b2358958c07eff64, 1cd87e1b16661b53378220deccb81e1b27aeade1, ae4b9bc91cd043ba205113d9d5fefb5ececf9910