The study introduces ICE (Iterative Correction and Eigenvector decomposition), a method to correct biases in Hi-C data and reveal chromosomal organization. Hi-C data is prone to systematic biases due to technical and biological factors, leading to uneven visibility of genomic regions. ICE corrects these biases by assuming equal visibility for all loci and iteratively refining contact maps. This process decomposes Hi-C data into a matrix of relative contact probabilities, enabling unbiased comparisons between datasets. The method involves aligning read-pairs from Hi-C data, filtering out molecular byproducts, and applying iterative correction to remove biases. The corrected maps are then decomposed into eigenvectors, which represent higher-order chromatin organization. The first eigenvector (E1) captures global chromatin features, showing strong correlations with GC content, replication timing, and histone marks. It also reflects the organization of chromatin compartments and is robust across experiments. The second and third eigenvectors (E2 and E3) capture local chromatin features, such as centromere and telomere enrichment. These eigenvectors reveal that chromosomal arms have distinct interaction patterns, with higher contact probabilities near centromeres and telomeres. The consistency of these patterns across human and mouse genomes suggests evolutionary conservation of chromosomal organization. The study also compares E1 and E2 with previously identified chromatin types, finding that E1 better captures epigenomic variation than the three chromatin types. Overall, ICE provides a robust framework for analyzing chromosomal organization at megabase resolution, revealing key features of higher-order chromatin structure. The method is applicable to various Hi-C datasets and can be used to compare chromosomal interactions across species and cell types. The results highlight the importance of iterative correction in uncovering the true spatial organization of chromosomes.The study introduces ICE (Iterative Correction and Eigenvector decomposition), a method to correct biases in Hi-C data and reveal chromosomal organization. Hi-C data is prone to systematic biases due to technical and biological factors, leading to uneven visibility of genomic regions. ICE corrects these biases by assuming equal visibility for all loci and iteratively refining contact maps. This process decomposes Hi-C data into a matrix of relative contact probabilities, enabling unbiased comparisons between datasets. The method involves aligning read-pairs from Hi-C data, filtering out molecular byproducts, and applying iterative correction to remove biases. The corrected maps are then decomposed into eigenvectors, which represent higher-order chromatin organization. The first eigenvector (E1) captures global chromatin features, showing strong correlations with GC content, replication timing, and histone marks. It also reflects the organization of chromatin compartments and is robust across experiments. The second and third eigenvectors (E2 and E3) capture local chromatin features, such as centromere and telomere enrichment. These eigenvectors reveal that chromosomal arms have distinct interaction patterns, with higher contact probabilities near centromeres and telomeres. The consistency of these patterns across human and mouse genomes suggests evolutionary conservation of chromosomal organization. The study also compares E1 and E2 with previously identified chromatin types, finding that E1 better captures epigenomic variation than the three chromatin types. Overall, ICE provides a robust framework for analyzing chromosomal organization at megabase resolution, revealing key features of higher-order chromatin structure. The method is applicable to various Hi-C datasets and can be used to compare chromosomal interactions across species and cell types. The results highlight the importance of iterative correction in uncovering the true spatial organization of chromosomes.