It's the 21st century version of "guns or butter?"
The project would determine the sequence of the DNA in at least 12,500 tumor samples, 250 samples from each of 50 major types of cancer. By comparing the order of the letters of the genetic code in the tumor samples with one another and with sequences in healthy tissue, it should be possible to pinpoint mutations responsible for cancer.
But the proposition is extremely daunting. In general, each tumor cell holds a full panoply of human DNA, a string of three billion letters of the genetic code. So determining the full sequence of all the tumors would be the equivalent of 12,500 human genome projects. At a cost of many millions of dollars for one genome, the full project would be out of the question for now.
So the cancer proposal for now is to sequence only the active genes in tumors, which make up 1 percent to 2 percent of the DNA. Even that would require at least 100 times as much sequencing as the Human Genome Project.