People with these skills are valuable to industry. "Pharma companies are desperately trying to organize and prioritize their drug targets," Sjölander says. "To do this you begin by making inferences about functions that are embodied in sequence."

However, that skill does not come easily. To develop fully the ability to make inferences, one needs to feel comfortable moving back and forth between computational and experimental data. "It's important for people who want to work in this field to have a true basis of understanding in more than one field. In addition to molecular biology, math, statistics and some elements of machine learning would be helpful," she says. "You can't do it with knowledge of just one specialty."

Sjölander thinks that bioinformatics should be a science in its own right. The demand certainly exists. The sheer quantity of data being generated from such a wide variety of sources requires the constant creation of new methods using mathematical or statistical modelling. "There's so much data being generated and we need to have methods that will respond to its particularities."

To increase the supply of computational biologists to deal with the demand being created by the data deluge, Sjölander believes that a cultural change is in order. "Historically, biologists haven't felt comfortable talking to mathematicians and computer scientists, who in turn don't really understand a lot of the issues in biology," she says. "That has to change."