A method of determining a probability that a human test subject has colorectal cancer as opposed to not having colorectal cancer is disclosed. The method comprises, for each gene of a set of one or more genes selected from the group consisting of ANXA3, CLEC4D, IL2RB, LMNB1, PRRG4, TNFAIP6 and VNN1: determining a level of RNA encoded by the gene in blood of the test subject, thereby generating test data; providing positive control data representing levels of RNA encoded by the gene in blood of human control subjects having colorectal cancer, and providing negative control data representing levels of RNA encoded by the gene in blood of human control subjects not having colorectal cancer; and determining a probability that the test data corresponds to the positive control data and not to the negative control data, where the probability that the test data corresponds to the positive control data and not to the negative control data represents the probability that the test subject has colorectal cancer as opposed to not having colorectal cancer.