LibMCS , Jarp & Ward?

Hi





i m using Version 3.2.3 of Libmcs and Jchem version is jchem.version=3.2.3 .





1) I tried to run the LibMCS program in command line with different options like fast search(-f), -n ,-l etc. But this is not reflectin g in the result.?


But these parameters are working in GUI.can u help me in this problem?


The input file i have used is also attaching.





2)in both the programs Jarp n ward, "central objects of clusters " are used ."they are A central object has the smallest sum of dissimilarities to the other objects in the cluster".


How can be useful for that for interpretation?"see the attached file, in tht center of objects are @, what is it actly means?





3)with the PF descriptor, both the programs giving different results(i mean clusters) i know since both r using different algorithms , the results will be


different, but how can be conclude from tht?





As a biginer the question can be silly, but please help me.

no text

achuz wrote:

Hi i m using Version 3.2.3 of Libmcs and Jchem version is jchem.version=3.2.3 . 1) I tried to run the LibMCS program in command line with different options like fast search(-f), -n ,-l etc. But this is not reflectin g in the result.? But these parameters are working in GUI.can u help me in this problem? The input file i have used is also attaching.

...snip





hi


I think i posted a link already yesterday, maybe it was deleted. I am quite sure that libmcs still does not allow


options in the commandline. see:





http://www.chemaxon.com/forum/ftopic2388.html&start=0&postdays=0&postorder=asc&highlight=








Tobias

Thank you Tobias....








cheers


Achuz

Quote:

1) I tried to run the LibMCS program in command line with different options like fast search(-f), -n ,-l etc. But this is not reflectin g in the result.? But these parameters are working in GUI.can u help me in this problem? The input file i have used is also attaching.

Tobias kindly quoted an earlier post which admits the the documentation and the software don't match... Apologies about the inconvenience, the new release should fix this issue.

Quote:

2)in both the programs Jarp n ward, "central objects of clusters " are used ."they are A central object has the smallest sum of dissimilarities to the other objects in the cluster". How can be useful for that for interpretation?"see the attached file, in tht center of objects are @, what is it actly means?

One can use this single structure to represent the entire cluster. Any other instance should do, but for many people this central one is more reasonable than a randomly chosen one.


@ is just a notation to identify the centroid of the cluster.


The centroid means that all other structures in the cluster are more similar to this particular one that to any other.


Strictly speaking the center of the cluster is a virtual point in a multidimensional space and the structure identified as "the centroid" is that element of the structure which is the nearest to the theoretical center. (Also, it was supposed that the cluster is a sphere. So the center of the cluster is the same as the center of the sphere.)


Does this help?

Quote:

3)with the PF descriptor, both the programs giving different results(i mean clusters) i know since both r using different algorithms , the results will be different, but how can be conclude from tht?

Indeed, different clustering methods result in different clusters. The selection of the most appropriate algorithm does depend on various circumstances. A primary factor is the size of the input.


Books and review on clustering should provide good guidance in selecting the most feasible method.





Regards,


Miklos

Thank you very much .. miklos