Deep sequencing technology provides efficient and economical production of large numbers of randomly positioned, relatively short, estimates of base identities in DNA molecules. Application of this technology to mRNA samples allows rapid examination of the molecular genetic environment in individual cells or tissues, the transcriptome. However, assembly of such short sequences into complete mRNA creates a challenge that limits the usefulness of the technology, particularly when no, or limited, genomic data is available. Several approaches to this problem have been developed, but there is still no general method to rapidly obtain an mRNA sequence from deep sequence data when a specific molecule, or family of molecules, are of interest. A frequent requirement is to identify specific mRNA molecules from tissues that are being investigated by methods such as electrophysiology, immunocytology and pharmacology. To be widely useful, any approach must be relatively simple to use in the laboratory by operators without extensive statistical or bioinformatics knowledge, and with readily available hardware.
An approach was developed that allows de novo assembly of individual mRNA sequences in two linked stages: sequence discovery and sequence completion. Both stages rely on computer assisted, Graphical User Interface (GUI)-guided, user interaction with the data, but proceed relatively efficiently once discovery is complete. The method grows a discovered sequence by repeated passes through the complete raw data in a series of steps, and is hence termed 'transcriptome walking'. All of the operations required for transcriptome analysis are combined in one program that presents a relatively simple user interface and runs on a standard desktop, or laptop computer, but takes advantage of multi-core processors, when available. Complete mRNA sequence identifications usually require less than 24 hours. This approach has already identified previously unknown mRNA sequences in two animal species that currently lack any significant genome or transcriptome data.
As deep sequencing data becomes more widely available, accessible methods for extracting useful sequence information in the biological or medical laboratory will be of increasing importance. The approach described here does not rely on detailed knowledge of bioinformatic algorithms, and allows users with basic knowledge of molecular biology and standard laboratory computing equipment, but limited software or bioinformatics experience, to extract complete gene sequences from deep-sequencing data.