Figure 3.

The pseudo code of the algorithm MBI. Input: A gene expression matrix A with m genes/rows and n samples/columns. Two parameters k₁ and k₂, where k₁ and k₂ are both positive integers and where k₁ is the number of partitions of the m genes and k₂ is the number of partitions of the n samples. Output: Two assignment matrices: Y₁* as the gene/row assignment matrix, Y₂* as the sample/column assignment matrix, and (k₁ × k₂) co-clusters of the matrix A. Main variables: A nonnegative integer k as the loop counter; A k₁ × k₂ matrix X with each entry a real number as the artificial central point of one of the k₁ × k₂ coclusters of the matrix A; A m × k₁ matrix Y₁ as the row assignment matrix with {0, 1} as the value of each entry; and A n × k₂ matrix Y₂ as the column assignment matrix with {0, 1} as the value of each entry.