Estimating the affinity of probes with unknown sequence. Sequence information was artificially removed from one out ever 100 genes from the Golden Spike data set and the corresponding probes' affinity categories were estimated from the data. There is a positive correlation between estimated and true categories, particularly for high-affinity probes.
Turro et al. BMC Bioinformatics 2007 8:439 doi:10.1186/1471-2105-8-439