The authors examined cytosine methylation using massively parallel sequencing on the Genome Analyzer. They developed bisulfite padlock (molecular inversion) probes (BSPP) to target CpG locations (not necessarily concentrated in CpG islands or promoter regions) in human B-lymphocytes, fibroblasts and induced pluripotent stem cells. They also used methyl-sensitive cut counting (MSCC) to generate genome-wide, nontargeted, data in the DNA of B-lymphocytes. In highly expressed genes there was high methylation in the gene-body with low methylation in the promoter region.

Investigation of single-nucleotide DNA methylation was carried out on arbitrary targets on chromosomes 12, 20, and 34. They used ~30,000 bisulfite padlock probes (BSPP) to study 2,020 CpG islands in human fibroblast and pluripotent stem cell lines and found slightly enhanced chromosome-wide cytosine methylation in the latter. Used Genome Analyzer for single-molecule bisulfite sequencing.

Investigation of microRNAs, which regulate gene expression, in white leghorn chicken embryo somites (connective tissue precursors) was carried out using the Genome Analyzer. The authors mapped 340,415 reads to the genome, which represented 1701 sequences. They found 85 known miRNAs and, after further investigation via Northern blot, they found eight novel miRNAs, one of which was at a high expression level in the somites.

The authors used high-throughput phenotypic analysis, on the Genomye Analzyer, to examine whole genome variations in two yeast strains—Saccharomyces cerevisiae and S. paradoxus. They reported most of the genome sequences and found 235,127 SNPs and 14,051 indels in S. cerevisiae, 623,287 SNPs, 25,267 indels in S. paradoxus, along with 38 novel hypothetical open reading frames. They elucidated parallels between phylogenetic relationships and phenotypic variation, with a far broader variation in S. paradoxus than S. cerevisiae.

Amplification artifacts can cause duplicated sequences which can weight read coverage and effect genome assembly and variation analysis. To counter this, the authors utilized (G+C) poor, neutral, or rich genomes from Plasmodium falciparum, Escherichia coli, and Bordetella pertussis, respectively. They developed a library preparation method without PCR but with fully ligated template strands being enriched by the cluster amplification step of the Genome Analyzer II system. This improved SNP detection, especially for P. falciparum, simplified assembly, and speeded up experimentation time.

RNA-seq using the Genome Analyzer helped examine transcriptional response of Burkholderia cenocepacia from soil or from cystic fibrosis sputum. A soil-like condition produced more gene expression in chromosomes 2 and 3, related to transcription, signal transduction, and nitrogen acquisition; a CF-like condition produced more on chromosome 1, related to virulence factors. More expansive gene expression in soil B. cenocepacia was proposed to be due to increased need for competition in this environment. They discovered 12 novel putative noncoding RNAs in the soil environment.

UV cross-linking and immunoprecipitation assay (CLIP) creates ribonuclease digested 40-80 nucleotide lengths of RNA with 3’ phosphate and 5’ hydroxyl groups. The 5’ end is phosphorylated by T4 poly-nucleotide kinase (PNK) to enable ligation. The author’s PNK does not have 3’ phosphatase activity, so, unlike some commercial PNK enzymes, RNAs cannot self-ligate. The novel protocol enabled this to be performed on beads. Prepared cDNA was sequenced using the Genome Analyzer II.

The authors developed a DNA-methylation assay (Methyl-seq) to examine over 90,000 regions using human embryonic stem cells (hESCs), their derivatives, and liver samples. In Methyl-seq, DNA is digested with methyl-sensitive restriction enzymes to enable creation of a fragment library for sequencing by the Genome Analyzer. They found differentiation resulted in 2-11% DNA methylation changes. hESC differentiation has both methylation (at transcription-related sites) and demethylation, with only the latter, at transcription-related sites, in fetal liver development.

The authors wanted to “define a transcriptome ab initio,” using the unannotated genome sequence of Saccharomyces cerevisiae, analyzed with short reads from massively parallel sequencing of cDNA (using the Genome Analyzer). Results, using novel algorithms, included defining 86% of expressed genes, finding 160 novel transcription units over 250bp, demarcating 86%/77% of 5’/3’ UTR boundaries, identifying 83% of known splice junctions, and finding 25 novel introns. The transcript absolute and relative expression level data was comparable to current microarray methods.

The Genome Analyzer II system was utilized to help create nucleosome maps of Saccharomyces cerevisiae and Candida albicans. Resultant gene sets were categorized into those closely associated (category I, growth-related), negatively associated (category II, stress-related), or species differentially associated (category III, aerobic respiration-related) with growth-related gene expression. They found promoters of category I genes encoded for open nucleosome organization, of category II, closed nucleosome organization, and in category III, C. albicans had open nucleosome-, and S. cerevisiae closed nucleosome-encoding aerobic respiration gene promoters.

Mouse models of medulloblastomas (MBs) were used to investigate micro RNA (miRNA) contribution to tumor development. Nine miRNAs encoded by the miR-17 approximately 92 cluster (Oncomir-1) were shown to be overexpressed in cerebellar granule neuron progenitors, which some MBs originate from, as well as developing mouse cerebellum, but not in mature neurons. The authors used “Illumina sequencers” to sequence small RNAs linked to specific initiating mutations from post-natal day 6 and 1 month old mice in MB models.

To investigate gene-expression in a colon cancer cell line, the authors used their own “oligo-capping method” and the Genome Analyzer to conduct a massively parallel sequencing study. They examined positional information of transcriptional start sites (TSSs) and gene expression with a comparison between cells in both normal and hypoxic conditions. They found genome-wide transcriptional alterations between the two conditions, including regions with enriched hypoxia-induced promoters. Their methods enabled “absolute quantification of the transcripts” without the need for probes or PCR primers.

Chromatin-immunoprecipitation sequencing (ChIP-seq) was used to investigate growth factor independent 1 (Gfi-1). This transcriptional repressor aids Th2 cell expansion and inhibits differentiation of naïve CD4 T-cells to Th17 or iTreg cells. Gfi-1 itself is repressed by TGF-beta. Th17 cells produce IL-17a and IL-17f , involved in immunity. Some iTreg CD4-cells express the immune response-suppressing CD103. The authors showed that Gfi-1-lysine-specific demethylase 1 repressive complex binds to the “intergenic region of II17a/II17f loci and to introns 1 of CD103.”

The authors developed immuno-SSAKE (iSSAKE) assembly software to investigate T-cell metagenomes utilizing the unique profile of the CDR3. They paired the alignment of sequenced short reads, constructed using the Genome Analyzer, and seeded de novo assembly. They created 20 million reads (36, 42, or 50 nt) with 1-2% random error, and seeded “iSSAKE de novo assemblies of CDR3.” The use of 42 nt and 50 nt reads to characterize sTCRß sequences was more sensitive than 36 nt reads.