Mutations in base pairs are like spelling or grammatical mistakes. In some cases they’re benign. In a 2016 study of seven genes in 102 patients, over 15,000 “variations of uncertain significance” were found. Other variations, however, do pose potential harm. These are known as “deleterious” mutations and can, in some instances, lead to cancer. It depends on how the body reads them and interacts with other genes.

Studies have led to the discovery of a wealth of such mutations. For example, the scientific community has known for decades that there are three relatively common variations in two genes, BRCA1 and BRCA2, that heighten the risk for breast and ovarian cancer.

In March of last year, the US Food and Drug Administration gave the company 23andMe approval to market its test for those BRCA mutations to the general public.

To understand how this test works, first consider the following paragraph from Moby Dick. The letters in red represent variations found in the gene.

It was just between daybreak and sunrise of a morning of the second day, when they were warshing down the decks, that a stupid Teneriffe man, drawing water in the main-chains, all at once shouted out, ‘There she rolls! there she rolls!’ Jesu, what a whale! It was Moby Drake.

23andMe’s test does not screen for every one of the over 1,000 mutations in each of the BRCA1 and BRCA2 genes that are linked with heightened breast or ovarian cancer risk. So while the appearance of the three variations that 23andMe does test for could help detect breast or ovarian cancer risk, their absence doesn't provide a conclusive diagnosis either. In the paragraph above, the red letters represent these three variations.

But here’s the same paragraph from above with all possible mutations:

That’s a fairly good representation of the amount of difference that 23andMe’s tests could miss.

Further, BRCA1 and BRCA2 are just two genes. To date, researchers have found thousands of potential markers for cancer risk in at least 36 different genes out the more than 20,000 that make up human DNA.

Establishing cancer risk by only analyzing the BRCA genes is like reading only one of 85 paragraphs from the entirety of Moby Dick.

DNA reports like those offered by 23andMe are even less accurate when looking at ancestry.

We inherit 50% of our genes from each biological parent. Since each subsequent generation inherits exactly half of the previous generation, establishing lineage from your DNA alone is difficult because it contains only fragments of history.

This would be true even if the test sequenced your entire genome. But DTC companies do not read your entire genome. Doing so is costly and time prohibitive. Instead, they sequence only bits and pieces.

To put it another way: if reading Moby Dick front to back is a complete decoding of your genome, then these ancestry tests would be like skimming—not even reading!—the CliffNotes.

These "notes" are compared to the test-provider’s genomic database, made up of DNA samples from its customers around the world. If the test finds similarities between your DNA fragments and the DNA fragments in its customer base, then it improves a company’s confidence in an ancestry match. But shards of DNA matches don’t prove ancestry. They simply show that your genes look, through one lens, similar to a collection of other people’s genes.

We can visualize the uncertainty this introduces by attempting to figure out Moby Dick’s “ancestry,” based on comparing sentences in the book with those in other works of fiction. In our experiment, every complete-sentence match increases the confidence level that Melville and the compared works of literature are related:

Ancestry report

Moby Dick











This analysis suggests a surprising level of similarity between the prose of Moby Dick and some of the touchstone works of Russian literature. But similarity does not prove ancestry. To truly understand the work’s influences, you’re better off simply reading the whole book.