search

PadiLab / CRANE

GNU General Public License v3.0
2 stars 0 forks source link

alpaca.crane #2

Open zenglongjin opened 2 years ago

zenglongjin commented 2 years ago

Dear Padi: We made the input file as described in ALPACA, but after we run it, we get the following error: Weights detected. Building condor object with weighted edges. Error in condor.cluster(ctrl.condor, project = F) : More than one connected component detected, method requires only one connected component

In addition, theoretically speaking, the number of regulatory networks of the two subtypes is different, so I don't know whether it is reasonable for us to directly take the common subset. Thank you very much for your attention and consideration.

meghapadi commented 2 years ago

Hi, ALPACA does assume that you are inputting two connected graphs. If your networks are not connected, one option is to input only the largest connected component of your network, if you are able to tolerate this partial loss of information. Another option may be to assign a very low "pseudo" edge weight to all the missing edges in your network. (We haven't tried this before ourselves.) For your second question, I don't understand what you mean by "number of regulatory networks". Do you mean the number of connected components is different for the two subtypes? Best, Megha

zenglongjin commented 2 years ago

Dear Padi: Your reply is pertinent, the number of connection points is different between subtypes and RGBM R package is the network construction algorithm for this hypothesis (PMID: 29361062). Our results show that the TF-Target with fewer intersections this will result in an error reported by alpaca.crane function and seems to be independent of the positive or negative weight. Thank you very much for your attention and consideration.

------------------ 原始邮件 ------------------ 发件人: "PadiLab/CRANE" @.>; 发送时间: 2022年1月1日(星期六) 凌晨5:01 @.>; @.**@.>; 主题: Re: [PadiLab/CRANE] alpaca.crane (Issue #2)

Hi, ALPACA does assume that you are inputting two connected graphs. If your networks are not connected, one option is to input only the largest connected component of your network, if you are able to tolerate this partial loss of information. Another option may be to assign a very low "pseudo" edge weight to all the missing edges in your network. (We haven't tried this before ourselves.) For your second question, I don't understand what you mean by "number of regulatory networks". Do you mean the number of connected components is different for the two subtypes? Best, Megha

— Reply to this email directly, view it on GitHub, or unsubscribe. Triage notifications on the go with GitHub Mobile for iOS or Android. You are receiving this because you authored the thread.Message ID: @.***>