Dear Amber Users,
I used the ptraj cluster command with the sieve option in order to cluster my
MD data (number of structures is 26371).
As described in the manual, I used the DBI as clustering algorithm in order to
obtain the metrics of the clustering.
My original command was as follows, which resulted in 4 clusters:
cluster out 1.00MC1_e2.3_sieve10 representative pdb average none all binpos
averagelinkage epsilon 2.3 mass rms sieve 10 xyzATOMS
After changing averagelinkage to DBI at the and of the output:
# Added by CLUSTER_DBI
#Cluster 0 has average-distance-to-centroid 1.250304
#Cluster 1 has average-distance-to-centroid 1.100415
#Cluster 2 has average-distance-to-centroid 1.432434
#Cluster 3 has average-distance-to-centroid 0.500635
#Cluster 4 has average-distance-to-centroid 1.310167
#Cluster 5 has average-distance-to-centroid -nan
#Cluster 6 has average-distance-to-centroid 0.983271
#Cluster 7 has average-distance-to-centroid 1.153958
#DBI: 2.83379
#pSF: 4052.40918
#SSR/SST: 0.51831
In this case the number of the cluster increases to 8. I obtained several
warnings as well:
Warning! Line length 20479 instead of 26371 found in cluster output file
1.00MC1_e2.3_sieve10.txt
Warning! Line length 5892 instead of 26371 found in cluster output file
1.00MC1_e2.3_sieve10.txt
.
.
Too many iterations in routine JACOBI
convergence failed
.
.
*** WARNING! No cluster centroid for point 20479
*** WARNING! No cluster centroid for point 20480
*** WARNING! No cluster centroid for point 20481
*** WARNING! No cluster centroid for point 20482
*** WARNING! No cluster centroid for point 20483
*** WARNING! No cluster centroid for point 20484
.
.
.
Can anyone suggest me how can I solve this discrepancy?
Thank you for your help in advance!
Balazs Jojart
_______________________________________________
AMBER mailing list
AMBER.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber
Received on Thu Mar 17 2011 - 05:30:02 PDT
