Compression Method

Processes | Genetics | Compression Method

Compression Method

Use this feature to select the algorithm for compressing the K matrix .

Option:

Select this Option to:

Optimized

Use the method proposed by Zhang et al. (2010) ¹ where the rows of the kinship matrix are iteratively clustered via SAS PROC CLUSTER across a range of compression levels.

At each level, a mixed model to the specified trait is fit (with the SNP effect excluded) with the compressed K matrix defining the covariance of the random effect . The final chosen compression level is that which optimizes the fit of the mixed model.

Automated

Compress the matrix via hierarchical clustering in PROC CLUSTER and PROC TREE based on a specified fixed number of clusters.

Interactive

Launch the JMP Hierarchical Clustering Platform. Compression level and cluster group membership can be decided visually.

Note : Compression via the Interactive mode is done through an action button that re-launches and updates the K Matrix Compression dialog to produce the compressed matrix by rerunning the process using the Automated method where Number of Clusters is decided from the JMP cluster platform. Be sure to save settings that you want to keep that use Interactive mode before implementing the action button.

1

Zhang, Z., E. Erzoz, et al. (2010). Mixed linear model approach adapted for genome-wide association studies. Nature Genetics 42: 355-360.

To Specify the Compression Method

Select the desired method by clicking within the appropriate radio button.

Option:	Select this Option to:
Optimized	Use the method proposed by Zhang et al. (2010) ¹ where the rows of the kinship matrix are iteratively clustered via SAS PROC CLUSTER across a range of compression levels. At each level, a mixed model to the specified trait is fit (with the SNP effect excluded) with the compressed K matrix defining the covariance of the random effect . The final chosen compression level is that which optimizes the fit of the mixed model.
Automated	Compress the matrix via hierarchical clustering in PROC CLUSTER and PROC TREE based on a specified fixed number of clusters.
Interactive	Launch the JMP Hierarchical Clustering Platform. Compression level and cluster group membership can be decided visually. Note : Compression via the Interactive mode is done through an action button that re-launches and updates the K Matrix Compression dialog to produce the compressed matrix by rerunning the process using the Automated method where Number of Clusters is decided from the JMP cluster platform. Be sure to save settings that you want to keep that use Interactive mode before implementing the action button.