Galaxy |

NSPDK_candidateClusters (version 9.2.3.1)

data_svector:

data_fasta:

data_names:

Multiple iterations:

for single iteration- NO, for multiple-YES

Number of current iteration :

maximum radius :

-R

maximum distance relations:

-D

Deactivate caching of kernel value computation (-no-cache):

to minimize memory usage

eccess neighbour size factor:

-ensf

Use shared neighbourhood to weight center density (-usn):

by default true

flag to output clusters (-oc):

by default true

Number of nearest neighbors:

-knn num

Number of hash functions :

-nhf num

Maximal number of hash functions :

Size of step for increasing hash functions :

The number of hash functions is increased by this value after each iteration.

Maxinum number of clusters :

What it does

Copmutes global feature index and returns top dense sets. The candidate clusters are chosen as the top ranking neighborhoods provided that the size of their overlap is below a specified threshold. For more information see Fast neighborhood subgraph pairwise distance kernel paper.

Parameters

-knn : <num nearest neighbors> (default: 10)
-otknn : flag to output true (i.e. implies full kernel matrix evaluation) k-nearest neighburs (default: 0)
-oc : flag to output clusters (default: 0)
-nhf : <num hash functions> for the Locality Sensitive Hashing function (default: 250)
-ensf : <eccess neighbour size factor> (default: 10) (0 to avoid trimming)
-usn : use shared neighbourhood to weight center density (default: 0)