Gene Expression Detection Assay for Cancer Clinical Use

Narrandes, Shavira; Xu, Wayne

Principle RNA sequencing (RNA-Seq), which directly sequences and counts the mRNA molecules in the whole transcriptome, can be selected to measure gene expression to detect genomic changes in disease states. Four steps are involved in RNA-Seq: the mRNA transcript is fragmented and primers are allowed to randomly anneal to mRNA segments; first and second strand cDNA synthesis from reverse transcription of the mRNA fragments result in double-stranded cDNA; the ends of the fragments are tagged with a phosphate group and poly(A) tail; and finally, adaptors are ligated onto the cDNA fragments so that PCR amplification and sequencing can be performed. 43 Sequencing involves the continual addition of fluorescently labelled nucleotide bases of four colours to the adaptor sequence, followed by photo imaging by a machine each time a base is added. These images are then translated into a sequence of bases. 44 , 45 The MiSeq® bench-top appliance sequences the cDNA templates and generates sequence reads in densities of 10 million individual molecular clusters per square centimeter through cluster generation, sequencing by synthesis (SBS), and data analysis. 46 Kit/RNA library preparation MiSeq RNA preparation kits, such as Illumina's Tru-Seq RNA-Seq kit, can be purchased from manufacturers. The kit includes reverse transcriptase for the reverse transcription of RNA into cDNA, primers, and adaptors. Sample preparation results in the ligation of adaptors to both ends of a nucleic acid fragment. 46, 47 Data analysis MiSeq conducts the basic data analysis with the MiSeq Control Software (MCS) and the Sequencing Analysis Viewer (SAV) installed on another computer. The MiSeq Reporter generates FASTQformat files containing the sequenced reads for each sample and their quality scores. The MiSeq Analysis folder contains a BaseCalls folder where the FASTQ files are sent to. The Targeted RNA workflow outputs a BAM file that contains the reference genome, with various annotations and set chromosome sizes, and a target hit file with aligned replicate counts per transcript. 47 , 48 The fastq files exported from MiSeq can be analysed alternatively. Read alignment poses as a complication for RNA-Seq as the reads are compared to a reference genome where exon-exon splice junctions exist in RNA transcripts but not DNA transcripts. RNA-Seq alignment tools that correct for this include GSNAP, MapSplice, RUM, STAR, and TopHat. 48, 49 MapSplice, for example, searches for exon-splice junctions in a sample mRNA transcriptome and then determines a likely alignment for each mRNA tag to a reference genome. In the first phase, the "tag alignment" phase, potential alignments of the mRNA tags to the reference genome are identified. The second phase, "splice interference phase", splice junctions appearing in one or more tags are further studied for their quality and diversity of alignments to output a splice significance score. Following read alignment, transcripts are formed from the mapped reads using computer software. 50 Programs such as Cufflinks and FluxCapacitator count the total number of reads mapping to transcripts to quantify gene expression. 51 Data between each sample can be normalized by reads per kilobase per million library reads (RPKM). Normalization on expression levels per replicate is performed based on the number of aligned reads. Variance estimation requires dividing the raw variance of all sample IDs by the square of the mean. Deriving the p-value for differe