We mapped histone H3 lysine 4 di- and trimethylation and lysine 9/14 acetylation across the nonrepetitive portions of human chromosomes 21 and 22 and compared patterns of lysine 4 dimethylation for several orthologous human and mouse loci. Both chromosomes show punctate sites enriched for modified histones. Sites showing trimethylation correlate with transcription starts, while those showing mainly dimethylation occur elsewhere in the vicinity of active genes. Punctate methylation patterns are also evident at the cytokine and IL-4 receptor loci. The Hox clusters present a strikingly different picture, with broad lysine 4-methylated regions that overlay multiple active genes. We suggest these regions represent active chromatin domains required for the maintenance of Hox gene expression. Methylation patterns at orthologous loci are strongly conserved between human and mouse even though many methylated sites do not show sequence conservation notably higher than background. This suggests that the DNA elements that direct the methylation represent only a small fraction of the region or lie at some distance from the site.

Sites of transcription of polyadenylated and nonpolyadenylated RNAs for 10 human chromosomes were mapped at 5-base pair resolution in eight cell lines. Unannotated, nonpolyadenylated transcripts comprise the major proportion of the transcriptional output of the human genome. Of all transcribed sequences, 19.4, 43.7, and 36.9% were observed to be polyadenylated, nonpolyadenylated, and bimorphic, respectively. Half of all transcribed sequences are found only in the nucleus and for the most part are unannotated. Overall, the transcribed portions of the human genome are predominantly composed of interlaced networks of both poly A+ and poly A- annotated transcripts and unannotated transcripts of unknown function. This organization has important implications for interpreting genotype-phenotype associations, regulation of gene expression, and the definition of a gene.

We have developed a robust algorithm for copy number analysis of the human genome using high-density oligonucleotide microarrays containing 116,204 single-nucleotide polymorphisms. The advantages of this algorithm include the improvement of signal-to-noise (S/N) ratios and the use of an optimized reference. The raw S/N ratios were improved by accounting for the length and GC content of the PCR products using quadratic regressions. The use of constitutional DNA, when available, gives the lowest SD values (0.16 F 0.03) and also enables allele-based copy number detection in cancer genomes, which can unmask otherwise concealed allelic imbalances. In

The cause of mental retardation in one-third to one-half of all affected individuals is unknown. Microscopically detectable chromosomal abnormalities are the most frequently recognized cause, but gain or loss of chromosomal segments that are too small to be seen by conventional cytogenetic analysis has been found to be another important cause. Array-based methods offer a practical means of performing a high-resolution survey of the entire genome for submicroscopic copy-number variants. We studied 100 children with idiopathic mental retardation and normal results of standard chromosomal analysis, by use of whole-genome sampling analysis with Affymetrix GeneChip Human Mapping 100K arrays. We found de novo deletions as small as 178 kb in eight cases, de novo duplications as small as 1.1 Mb in two cases, and unsuspected mosaic trisomy 9 in another case. This technology can detect at least twice as many potentially pathogenic de novo copy-number variants as conventional cytogenetic analysis can in people with mental retardation.

Formalin-fixed, paraffin-embedded (FFPE) material tends to yield degraded DNA and is thus suboptimal for use in many downstream applications. We describe an integrated analysis of genotype, loss of heterozygosity (LOH), and copy number for DNA derived from FFPE tissues using oligonucleotide microarrays containing over 500K single nucleotide polymorphisms. A prequalifying PCR test predicted the performance of FFPE DNA on the microarrays better than age of FFPE sample. Although genotyping efficiency and reliability were reduced for FFPE DNA when compared with fresh samples, closer examination revealed methods to improve performance at the expense of variable reduction in resolution. Important steps were also identified that enable equivalent copy number and LOH profiles from paired FFPE and fresh frozen tumor samples. In conclusion, we have shown that the Mapping 500K arrays can be used with FFPE-derived samples to produce genotype, copy number, and LOH predictions, and we provide guidelines and suggestions for application of these samples to this integrated system.