HyFeaL

A fast and accurate dimension reduction framework for methylation microarray data analysis using hybrid feature learning.

The overview of HyFeaL

Computational pipeline

1) The preprocess instructions of Infinium Human Methylation 450k data is available at our previous work HyDML.

2) Methylation array data analysis with HyFeaL

Since the original Infinium 450k array data is too large, we upload test data for validating HyFeaL at Test.zip.

2-1: Input the preprocessed array data with size of n * m array, n is the number of the samples, m is the number of CpGs(features).

   Sample_id  cg_1   cg_2                cg_m

   1      0.15   0.75                     0.95

   2      0.85   0.12                     0.25

   ...        ...              ...

   n      0.35   0.45                     0.75

2-2: Implement HyFeaL to identify the robust DMS.

   id_comb1 = HyFeaL_1s(X,y,Q1=0.05)
   
   id_s1 = HyFeaL_2s(X,y,Q2=0.2,method='chi_square')
   
   id_s2 = HyFeaL_2s(X,y,Q2=0.2,method='fisher')
   
   id_s3 = HyFeaL_2s(X,y,Q2=0.2,method='f_score')
   
   id_s4 = HyFeaL_2s(X,y,Q2=0.2,method='reliefF')
   
   id_comb2 = HyFeaL_3s(id_s1,id_s2,id_s3,id_s4)
   
   X_fs = X[:,id_comb2]

2-3: Implement HyFeaL for visualization.

   X_2d = SGE_tsne(X_fs,y,perplexity=25)

Contact

[email protected]