This paper presents a clustering algorithm for multi-condition gene expression patterns. The algorithm is based on a stochastic model of gene expression data and is designed to recover cluster structures with high probability. The algorithm runs in O(n(log(n))^c) time, where n is the number of genes and c is a constant depending on the cluster structure and the error probability. The algorithm is also applied to real gene expression data and shown to perform well. The paper also introduces a practical heuristic approach, called CAST, which is based on the same ideas as the theoretical algorithm. CAST uses the similarity between unassigned vertices and the current cluster seed to make its next decision. The algorithm is tested on simulated data and shown to recover hidden cluster structures effectively. The paper also discusses the application of the algorithm to biological data, including temporal gene expression patterns and multi-experiment analysis. The results show that the algorithm can effectively cluster genes based on their expression patterns and that the clusters can be validated against known biological functions. The paper concludes that the algorithm is a useful tool for analyzing gene expression data and that further research is needed to improve its performance and applicability.This paper presents a clustering algorithm for multi-condition gene expression patterns. The algorithm is based on a stochastic model of gene expression data and is designed to recover cluster structures with high probability. The algorithm runs in O(n(log(n))^c) time, where n is the number of genes and c is a constant depending on the cluster structure and the error probability. The algorithm is also applied to real gene expression data and shown to perform well. The paper also introduces a practical heuristic approach, called CAST, which is based on the same ideas as the theoretical algorithm. CAST uses the similarity between unassigned vertices and the current cluster seed to make its next decision. The algorithm is tested on simulated data and shown to recover hidden cluster structures effectively. The paper also discusses the application of the algorithm to biological data, including temporal gene expression patterns and multi-experiment analysis. The results show that the algorithm can effectively cluster genes based on their expression patterns and that the clusters can be validated against known biological functions. The paper concludes that the algorithm is a useful tool for analyzing gene expression data and that further research is needed to improve its performance and applicability.