We prove the W[2]-completeness of the feature subset selection
problem when the cardinality of the subset is the parameter.
Aside from the many applications the problem has in the
data mining literature, the problem is highly relevant in
Computational Biology since it arises in differential gene
expression analysis using microarray technologies. It is also
related to genetic-based prognosis and regulatory interaction
discovery using DNA chip technologies.