For a description of the different classes of antidepressants, see the Wikipedia entry.
This article reports that SNPs can be used to predict whether norepinephrine reuptake inhibitors (NRIs), such as bupropion, or selective serotinin reuptake inhibitors (SSRIs, such as Paxil) will be more effective for a particular late-life major depression patient.
Common examples of tricyclic antidepressants are imipramine (brand names Tofranil, Janimine), amitriptyline, nortriptyline (Aventyl, Pamelor), desipramine (Norpramin, Pertofrane), and clomipramine (Anafranil). Tricyclic antidepressants are poorly metabolized by individuals homozygous for poorly metabolizing CYP2D6 alleles (such as CYP2D6*4, i.e. SNP rs3892097), who will therefore generally do better at lower doses of these drugs. [PMID 18070221]
The most important class of enzymes for the metabolism of most drugs are the cytochrome P450s. In general, a patient who is a poor metabolizer for a CYP known to metabolize a given drug may be at higher risk for adverse side effects from that drug, and may derive benefit from being prescribed lower doses of that drug, or alternatively, may benefit more from being prescribed a comparable drug metabolized by a different CYP for which that patient is a normal metabolizer.
For antidepressant drugs, an overview (adapted from Wall CA et al., Prim. Psychiatry 17(5), 2010) of which CYP is the primary metabolizer (shown as 1), or a secondary/substantial metabolizer (2), or is quite minimally involved (0), is shown in the following table:
SNPs in genes other than the CYP450s that influence how well a patient responds to certain antidepressants include: