Source Estimation and Connectivity with missing channels

Hey there,
I conducted a perception experiment with two conditions with a 64 channels EEG system. I did the preprocessing in eeglab and removed bad channels for all 31 subjects in the process. Afterwards I switched to brainstorm with the plan to compute the sources, followed by functional connectivity and finally compare the two conditions.
I encountered a problem regarding the missing channels.
I did not interpolate any channels which led to all the subjects having different numbers of channels. I tried to average both conditions but due to the different amount of channels that does not work.
Do I have to interpolate the missing ones?
I saw that there is an option that allows me to uniform list of channels. But I am not sure if any of the options there is favourable for my analyses.

I also tried to compute connectivity for all my subjects at once leading to a message that tells me that I have to compute it for each subject separately (with the reference to scripting).
So in addition to the question about the missing channels I'd be grateful for a hint on the correct order of steps to compute connectivity. Do I compute it for each subject separately or do I average all trials of one condition for all subjects and compute the connectivity for that averaged file?

Best wishes

A ways is to put all your recording (EEG) files in the Process1 tab and use the process: Standardize > Uniform list of channels with the option Keep all the channel names. This will find all the channels in all the files (which should add to the acquired 64 channels, unless a channel was missing for all subjects), then add those missing channels to the files that need them and mark them as bad.

Now that all the recording files have the same number of channels is possible to average them. The bad channels will not be taken into account to compute the average.

It is not recommended to compute connectivity on averaged files (nor subject nor group averaged). You could compute connectivity metrics per subject, and then perform group analysis.