The art and science of (indirectly) determining the (comparative) level of expression of the genes (plural) in a given state of a given system. Usually refers to attempts to determine a slice of the transcriptome of interest (ie. the examination of mRNA levels; often by reverse-transcribing the mRNA to cDNA and working with that instead - see below), but proteome-based approaches also exist.


mRNA Expression Analysis

If you wish to be contrary and use the term 'gene expression analysis' to refer to fishing expeditions for the presence or abundance of a single gene/allele or a mere handful of genes/alleles, then by all means hallucinate that here lie references to Southern blotting, fluorescently-labelled non positionally-encoded microbead arrays resplete with palindromic quenching tags, etc.

Survey methods
These are random sampling of cDNA libraries (and the use of ESTs) and its extension, Serial Analysis of Gene Expression (SAGE).

Essentially, we are randomly sampling mRNA populations, sequencing our samples, identifying the gene responsible, and assuming that mRNA levels in our sample correlate with the actual relative expression levels. Note that the sequence or SAGE tag must be in a database to identify the relevant gene.

These methods return a faithful representation of mRNA abundance (because they are statistically robust). Unfortunately, the sequencing effort required to detect (rare) mRNAs is prohibitive, even with SAGE. Furthermore, we cannot readily study changes in mRNA expression levels under different conditions (as we can with array methods, below) because we require a cDNA library as our basis.

Array methods
These include cDNAs on filters/membranes or glass slides (DNA arrays), and oligonucleotides on silica (DNA/oligonucleotide microarrays/chips).

Essentially, we are spotting known (ie. from a library) single-stranded cDNAs or oligonucleotides on a surface (remembering which position corresponds to which cDNA spot), incubating this surface with our mRNA population (to which we have have attached fluorescent tags), and taking a CCD image to identify which cDNAs have mRNA bound to them and how much.

These methods are fast and convenient, and we can assay cells from tissues/states for which cDNA libraries do not exist (by using that of a related organism). However, accuracy issues exist (and are exacerbated by the minimal duplications performed) and we cannot detect mRNAs that do not bind to the cDNAs or oligonucleotides that we chose to spot our surface with.

In the above arrays, specific cDNAs are identified by position. There is a recent movement towards experimenting with arrays with non-positionally encoded elements, using tagged silica microbeads held in suspension or distributed randomly on a surface (microbead arrays). Binding of target is still detected via fluorescence, but identity is now revealed through interrogation of ā€˜optical barcodesā€™ (unique ratios of two or more fluorescent dyes absorbed into each variety of bead) via flow cytometry (FACS) or multispectral imaging (CCDs). MPSS (see below) can also be considered among this brethen.

MPSS
A synthesis of the two paradigms is achieved in the Massively Parallel Signal Sequencing (MPSS) technique of Sydney Brenner (et al). This can be considered an extension of SAGE that avoids conventional sequencing.


Protein Expression Analysis

If you wish to be contrary and use the term 'gene expression analysis' to refer to fishing expeditions for the presence or abundance of a single protein or a mere handful of proteins, then by all means hallucinate that here lie references to Northern blotting, fusion proteins containing GFP, etc.

Techniques for the identification and quantification of the proteins present in a sample include two-dimensional gel electrophoresis, mass spectrometry (MALDI-TOF or electrospray), and the sequencing of (subpicomole amounts of) protein. Not my field. Yet. Spank me.