Hi Nina,
1, These numbers provide some key figures of your cluster hierachy.
I had 214 compounds in my library & my report is Level count = 5, Top level cluster count = 62 and total cluster count = 356.
|
|
If there are 5 levels in total, that means that clusters were formed in 4 consequtive steps. The bottom level of the cluster hierachy contains your 214 input compounds from your library. The next level groups some of these into initial cluster, the third level groups the initial clusters into larger cluster and so on until the top level (which the 5th in your case) has been formed. On this level there 62 clusters. This means that your 214 compunds were finally grouped into 62 clusters.
The total cluster count (which was 356 in your case) includes these 62 top level clusters as well as all internal clusters in levels 2 to 4. If you take a look at the tree-like display panel on top of the libmcs window then these concepts (like hierarchy, tree, clsuters etc.) can be visually captured.
2, With "Add one level" you can continue clustering, that is, add a 6th level, a new top level to the hierachy, in which your 62 clusters are merged even further. However, to get there you have to loosen the cluster termination conditions (i.e. when does the clustering process stop). This can be done in the Cluster menu, select Options from among the menu items, and change the Minimal MCS size parameter from 9 (the default value) to any smaller number. You can decrease it to 8, then Add one level, then if you aren't yet satisfied with your clusters go down to 7 and repeat this as many times as needed. IMO there's no chemical sense to go under 5.
3, Minimal MCS size in options refers to the smalles size of the maximum common substructure searched for by the algorithm. If this is set to 6 for instance, then clustering stop if the size of the scaffolds in the clusters reaches this size limit.
4, Maximum level count refers to the number of levels allowed in the cluster hierarchy. If this is set to a smaller number, then fewer steps are done, clustering terminates faster, larger number of clusters are formed and scaffolds in the clusters are also larger. In contrast to this, it the number is increased then clustering proceeds further thus on the top level fewer clusters with smaller scaffolds are formed.
Maximum allowed level count, Minimal MCS size are the termination conditions for the clustering process. If any of these limits are reached by the clustering process then the algorithm terminates and the result (the cluster hierarchy) is diplayed.
Does this help?
Regards
Miklos