A new estimate puts the paltry number of human genes more than 30 per cent higher. At around 30,000, the gene number estimated from the human genome sequence seemed scanty compared with the mustard weeds' (25,000) and the fruitfly's (13,000). Michael Cooke and colleagues at the Genomics Institute of the Novartis Research Foundation in San Diego, California compared the two gene sets and found that they overlap only by about half.
The researchers detect the activity of some of the genes in human tissues confirming that both groups predicted real genes. If the rest of the mooted genes are genuine, an estimate of 42,000 or more is closer to the mark.
According to the researchers, gene-prediction algorithms are at fault, rather than the underlying sequence itself. Cooke says that computational methods, which look for similarities to known gene sequences, are difficult to get right. The final number will remain unknown until firm biological evidence, such as gene activity, confirms each gene's existence. Such an analysis identified an extra 1,000 genes in fruitfly genome.