First, you might wish to “filter” your peak lists. This action compares each of your peak lists to the selected reference peak list and jettisons peaks that do not exhibit amide 1H and 15N frequencies found in the reference list. For the experiments handled by RunAbout, all spectra should look like HNCO spectra when projected on to an 1H-15N plane, so choose the hnco peak list as your “reference peak list” unless you have a special reason for not doing so. When filtering, RunAbout looks for reference peaks that are within two times the tolerance value set for the peak list. This increased tolerance is to minimize the probability of erroneosly removing peaks that are valid, but not well aligned.

Click the "filter" button. You should see something like this result in the SWKcon window:

Next, consider the “trim” function. For a given protein and a certain spectrum type, one expects a certain number of crosspeaks. If you picked your peaks conservatively, you should probably have more than the expected number of peaks, even after filtering. The “Trim” function allows you to cut out the less intense peaks, thus the ones most likely to not be valid, based on the number of expected peaks. Here, you should provide a “trim number”, which RunAbout equates, just for this purpose, with the number of residues in your protein; to be safe, give an estimate that’s ~10% larger than the number of residues in your protein. Ubiquitin bears 76 residues, so here the trim number is set to 84. Click “Trim”, and you should see something like this appear in your SWKcon window:

	Save, Now and Often
	Trimming could be considered somewhat more likely to remove "real" peaks than the above Filtering action. With any of these commands, and periodically throughout the assignment process, you should save versions of the project into a STAR file.

After filtering and trimming, click “Compress and Degap” to remove the filtered trimmed peaks from your peaklists, and renumber to them to remove numerical gaps in their sequence. This action will not result in any message window or text in the swkcon window.

Now you need to group your peaks into clusters. This will group together peaks from your peaklists that have the same amide H and N frequencies. Clustering is thus the basis by which sequential assignment is made: clustering determines which resonances in different peak lists belong to the same residues, and also begins distinguishing intra- and inter-residue peaks. Click "cluster Peaks" to group them on the basis of your reference peak list; you should see the following in the swkcon window:

Peaks are clustered together based on the tolerance values set up for each list. Additionally, the clustering is constrained so that every cluster must have one, and only one, peak from the reference list. Clusters will be assigned an identification number that corresponds to the index of that cluster's reference peak.

Some spurious clusters will probably be created in this process. The ones most easily identified as being misleading are those with just one or two peaks in the cluster - so-called "lonely clusters." You can easily get rid of these by setting a "lonely Limit," which corresponds to the minimum number of peaks in a legitimate cluster, and clicking "Purge Lonely". Here, we set the lonely limit to 2 so that all clusters with just one peak are eliminated, and all those with two or more are kept. After clicking "Purge Lonely", you should observe something like this in your swkcon window:

Note that this is just the tail end of a message that also announces how many lonely clusters were encountered.

The "Purge Lonely" button also renumbers the clusters, much like the "Compres and Degap" function renumbers peaks. If you originally had 84 clusters, for example, and five lonely clusters were purged, the resulting 79 clusters are numbered 0-78.

The actions panel has a button labeled "Trim All". Unlike the previous trim button, this one eliminates the peaks from clusters, and it does so by eliminating the peaks least likely to be included in the cluster so that the number of peaks remaining in the cluster matches the number expected. This action is normally performed in the next step, "Edit Clusters", on a cluster-by-cluster basis. But if you have already worked with this dataset before, you can use 'Trim All" to speed up your process; you should still check to see whether it eliminated more peaks that it should have. If you click 'Trim All", you should see something like this in your swkcon window: