search

Aleksobrad / single-cell-rcc-pipeline

Data files and code for analysis of single-cell ccRCC data for the manuscript "Tumor-Specific Cell Populations in Clear Cell Renal Carcinoma Associated with Clinical Outcome Identified Using Single-Cell Protein Activity Inference." Includes code for VIPER protein activity inference pipeline
35 stars 23 forks source link

Why is there no "Signaling Proteins" regulator set in Modules/ARACNe/ directory for mouse? #6

Open baoyulinn opened 1 year ago

baoyulinn commented 1 year ago

I can not find this "signaling protein" regulator set of mouse in this directory.

Does this mean that the regulator set is not needed? If this set is needed, do I need to create this regulator set by myself?

Thank you！

Aleksobrad commented 1 year ago

This set was needed to generate the ARACNe networks, which are pre-generated and included here, as generating these networks is the most time-intensive and computationally expensive step of the analysis.

If you wish to generate your own ARACNe networks I would either create your own regulatory protein lists or you can use the lists available at: https://github.com/califano-lab/PISCES/tree/master/data

On Thu, Sep 15, 2022 at 1:58 AM baoyulinn @.***> wrote:

I can not find this "signaling protein" regulator set of mouse in this directory.

Does this mean that the regulator set is not needed? If this set is needed, do I need to create this regulator set by myself?

Thank you！

— Reply to this email directly, view it on GitHub https://github.com/Aleksobrad/single-cell-rcc-pipeline/issues/6, or unsubscribe https://github.com/notifications/unsubscribe-auth/ABLHK2XHTBZODF2RY3CAY6LV6K3JJANCNFSM6AAAAAAQNBI5XU . You are receiving this because you are subscribed to this thread.Message ID: @.***>

baoyulinn commented 1 year ago

Thank you for your valuable advice!

This list (available at: https://github.com/califano-lab/PISCES/tree/master/data) still do not have "signaling protein" set. There are only three other sets (tf, cotf, surface protein).

image

Can i use these three regulator set (tf, cotf, surface protein) to generate ARACNe networks of mouse?

If the "Signaling protein" regulator sets is necessary like human's, can i use this methods to create this "signaling protein" regulator set? image

Aleksobrad commented 1 year ago

For analysis of mouse data in the past we have typically used only these three tf, cotf, and surface protein sets. An additional set can of course be added if a protein of interest to your research is not included in the sets that are currently collected.

On Thu, Sep 15, 2022 at 7:31 AM baoyulinn @.***> wrote:

Thank you for your valuable advice!

This list (available at: https://github.com/califano-lab/PISCES/tree/master/data) still do not have "signaling protein" set. There are only three other sets (tf, cotf, surface protein).

[image: image] https://user-images.githubusercontent.com/111479371/190388915-539bbd8c-49b0-4495-ba87-d473f07b1a4a.png

Can i use these three regulator set (tf, cotf, surface protein) to generate ARACNe networks of mouse?

If the "Signaling protein" regulator sets is necessary like human's, can i use this methods to create this "signaling protein" regulator set? [image: image] https://user-images.githubusercontent.com/111479371/190392001-3169f201-d7f9-4ec4-980c-0771daf997a2.png

— Reply to this email directly, view it on GitHub https://github.com/Aleksobrad/single-cell-rcc-pipeline/issues/6#issuecomment-1247976617, or unsubscribe https://github.com/notifications/unsubscribe-auth/ABLHK2VK4YOVFHWUXSRV4ILV6MCIBANCNFSM6AAAAAAQNBI5XU . You are receiving this because you commented.Message ID: @.***>

baoyulinn commented 1 year ago

Thank you very much for your answer!

baoyulinn commented 1 year ago

The article indicates screening for signaling proteins present in the intracellular and plasma membranes.

3455 signaling pathway related genes (annotated in GO biological process database as GO:0007165, ‘‘signal transduction’’ and in GO cellular component database as GO:0005622, ‘‘intracellular’’ or GO:0005886,‘‘plasma membrane’’)

Can this set of "signaling protein" regulators include extracellular cellular components? In other words can it include secreted proteins?

Aleksobrad commented 1 year ago

This protein set should NOT include secreted proteins— conceptually VIPER relies on downstream transcriptional effects to infer activity of upstream proteins. Secreted proteins are typically not transcriptionally upstream of anything in the same cell, and so their VIPER activity is not conceptually well-defined.

On Sun, Sep 18, 2022 at 4:56 AM baoyulinn @.***> wrote:

The article indicates screening for signaling proteins present in the intracellular and plasma membranes.

3455 signaling pathway related genes (annotated in GO biological process database as GO:0007165, ‘‘signal transduction’’ and in GO cellular component database as GO:0005622, ‘‘intracellular’’ or GO:0005886,‘‘plasma membrane’’)

Can this set of "signaling protein" regulators include extracellular cellular components? In other words can it include secreted proteins?

— Reply to this email directly, view it on GitHub https://github.com/Aleksobrad/single-cell-rcc-pipeline/issues/6#issuecomment-1250224543, or unsubscribe https://github.com/notifications/unsubscribe-auth/ABLHK2UQEDO3SE7YOGMMTI3V63KNTANCNFSM6AAAAAAQNBI5XU . You are receiving this because you commented.Message ID: @.***>

baoyulinn commented 1 year ago

I really appreciated your help and patience. Thank you so much for carving out time to answer my questions.

baoyulinn commented 1 year ago

In this step, the codes are to do a patient-level VIPER analysis. However, the nets made by your code are created for all cd45neg patients.

image

Do we need use all patients' net to do a patient-level VIPER analysis for each patient? If we want to do a patient-level VIPER analysis, shouldn't we use the patient's own corresponding net？

In addition, the function sil_subsample_v2_viper may not be provided in your code

image

I hope you will answer, thanks!

Aleksobrad commented 1 year ago

We generate the ARACNe networks at patient-specific level, and then run VIPER using all the networks generated from each patient. The benefit of this is that missing proteins where the ARACNe net only inferred a regulon in one patient can be filled in using networks generated from other patients. The best-matched network is automatically prioritized within the VIPER algorithm for inference of protein activity in a given sample, and this will typically be the network derived from that patient, but including all other networks allows for "filling in the gaps" wherever network data for a certain protein are lower-quality or missing in one patient compared to other patients.

Apologies for the exclusion of sil_subsample_v2_viper function and thank you for catching this-- that function can be completely replaced by the function sil_subsample() as we have subsequently tested and found that there is no difference or need for a viper-specific function here.

On Wed, Sep 21, 2022 at 10:53 AM baoyulinn @.***> wrote:

In this step, the codes are to do a patient-level VIPER analysis. However, the nets made by your code are created for all cd45neg patients.

[image: image] https://user-images.githubusercontent.com/111479371/191533915-4ae07f85-17c8-4e95-a658-194b5ccbfcf5.png

Do we need use all patients' net to do a patient-level VIPER analysis for each patient? If we want to do a patient-level VIPER analysis, shouldn't we use the patient's own corresponding net？

In addition, the function sil_subsample_v2_viper may not be provided in your code [image: image] https://user-images.githubusercontent.com/111479371/191537351-6ed7a55e-a45c-44bd-93b0-8e1602c08df8.png

I hope you will answer, thanks!

— Reply to this email directly, view it on GitHub https://github.com/Aleksobrad/single-cell-rcc-pipeline/issues/6#issuecomment-1253828129, or unsubscribe https://github.com/notifications/unsubscribe-auth/ABLHK2UKI5BSJQQMGFIBUKTV7MOPHANCNFSM6AAAAAAQNBI5XU . You are receiving this because you commented.Message ID: @.***>

baoyulinn commented 1 year ago

When I try to generate an ARACNe network, I always get some warnings (WARNING! NOT CONVERGENT! number of iterations= 1000) I would like to know if the result does not converge, can it be used in the next step?

image
Aleksobrad commented 1 year ago

We sometimes get that error with low-quality data— unsure exactly how to best address as it is fairly application-specific. Could try increasing the number of iterations or accepting the caveat of a few lower-quality networks if you also have good-quality networks.

On Fri, Sep 23, 2022 at 7:41 AM baoyulinn @.***> wrote:

When I try to generate an ARACNe network, I always get some warnings (WARNING! NOT CONVERGENT! number of iterations= 1000) I would like to know if the result does not converge, can it be used in the next step?

[image: image] https://user-images.githubusercontent.com/111479371/191952744-aa5552d2-10a4-4094-abd0-57f9c2bf36e4.png

— Reply to this email directly, view it on GitHub https://github.com/Aleksobrad/single-cell-rcc-pipeline/issues/6#issuecomment-1256105894, or unsubscribe https://github.com/notifications/unsubscribe-auth/ABLHK2S4JBA77FZQFM3TX5TV7WJPVANCNFSM6AAAAAAQNBI5XU . You are receiving this because you commented.Message ID: @.***>