**5. Future directions**

Though useful in predicting quality, AC, GT, and GC do not paint the whole picture of rice quality. Varieties identical in these three traits may be grouped into one quality class based on these parameters but consumers easily distinguish a premium variety from a lower quality one (Champagne et al., 2010). This could lead to low rates of adoption of newly developed improved varieties by farmers. Unfortunately, consumers are rarely able to describe the difference when they eat supposedly identical rice varieties (based on AC, GT, and GC if applicable). Hence, the next steps in discovering genes for sensory quality include finding descriptors for the sensory experience and developing phenotyping tools that can be used to quantify these descriptors. Once the phenotype is known, association mapping can begin, using appropriate populations. Such an approach will then lead to the delivery of validated genotyping tools to breeding programs.

An example of the value and need for phenotyping tools is the trait of aroma. Aroma is easy to define as present or absent in cooked rice. It was therefore possible to develop a phenotyping tool, in this case gas chromatography (Bergman et al., 2000), and then use mapping populations, genome-wide genotyping and sequencing of candidate loci to find genes and allelic variation (Kovach et al., 2009). Unfortunately, other sensory properties of rice are more abstruse because they are not as easily described by consumers. To find adjectives for these other sensory properties, descriptive sensory profiling is employed; a trained sensory panel assesses food for aroma, texture, and flavour (Champagne et al., 2010). Comparisons by a trained panel between similar varieties (in terms of AC, GT) but classed as premium and second-best showed that slickness, roughness, and springiness were textural attributes that separated the two classes while sweet taste, popcorn flavour, and metallic mouthfeel were the flavour attributes (Champagne et al., 2010). Without phenotyping tools, these traits could not be associated with genetic loci. Thus, discovering novel sensory quality genes goes in tandem with developing phenotyping tools.
