How to find a needle in a very big haystack

I have two sons. One is 25, the other 18. The 25-year-old is probably OK now. Bright and sensible, we would be extremely unlucky if serious harm were to befall him. For the younger one, though, there remains some cause for concern. Though equally bright and sensible, he is yet to live through the main age of risk for developing the most common and serious illness to affect young people. By and large they are a healthy bunch and serious illnesses are very rare. But schizophrenia is not rare at all.

Upwards of half a million people in Britain will develop schizophrenia at some point, mostly between the ages of 18 and 25. I've met hundreds of them myself. And each time I do I am aware not only of the patient in front of me, suffering terrible psychotic symptoms or confused and rambling, but also of two parents who once, like me, had a lively, promising child but who now face a future as "carers" of an adult with severe mental illness.

Although factors such as stress and cannabis use can increase the risk of schizophrenia it remains primarily a genetic disorder. We have known for years that some variations in the genetic code must profoundly impact this risk, and for years we have been trying to find them. Lately, we’ve made some progress. A recent paper in the British Journal of Psychiatry listed half a dozen deletions or duplications of large chromosomal segments which had been seen only in patients with schizophrenia and never in normal controls. Thus, they could be viewed as “causing” the illness. However, such large variants between them account for only 1 in 50 cases of schizophrenia and each one stretches over such a long region of a chromosome that it involves dozens of genes. We don’t know which of these genes is the critical one so we can’t tell which cellular system is being disrupted to cause the disease. Although these variants have the potential to provide an occasional patient with an explanation of why they have developed the illness, they do little towards moving us forward in understanding the fundamental abnormalities of brain functioning which would give us a better grounding to develop improved approaches for treatment and prevention.

To do that we need to identify some of the small variants, affecting just single DNA bases within individual genes, which together account for the bulk of the genetic contribution to risk. And to do that we need the sequences of all the genes from a large number of patients with schizophrenia and of controls. That way, we can begin to look for variants which are commoner in patients and we can then explore their effects in the cultures of nerve cells which we grow in our lab.

A couple of months ago, we got just what we’d been waiting for. Scientists at the Broad Institute had sequenced all 20,000 genes of 2,500 patients with schizophrenia and 2,500 controls. And, having published their own preliminary analysis, they made the sequence files available for others to download and carry out their own searches. Their preliminary analysis confirmed that rare variants did seem to congregate in particular genes plausibly involved in brain function. However no individual variant was common enough to be definitively assigned a role.

The problem is, there are an awful lot of variants. Ignoring “junk” DNA, I have around 25,000 variants in my genes. So do you. And we don’t have the same ones. In fact, I’ve probably got a couple of thousand that have never been noted to occur in anybody else. So if one of us has schizophrenia and the other one doesn’t it’s not going to be possible to say which of those thousands of variants are producing that effect. Even with sample sizes of 2,500 cases we can’t be sure – those 5,000 subjects have half a million variants between them.

And so we spend our days, evenings and weekends working on the problem. We write thousands of lines of computer code to analyse the data. To look for variants which are commoner in cases. To look for variants which are inherited in families along with disease. We compare variants to the 3 billion base sequence of the “normal” human genome to work out what amino acid change would be produced and whether this would be likely to disrupt the functioning of the protein for which the gene codes. The programs run for hours, or days. We look up genes in online databases to see whether they are known to be involved in some aspect of brain cell functioning or development which could plausibly affect the risk of schizophrenia. When we find one which seems promising, we genotype it in our own samples of thousands of patients and controls, painstakingly collected over the years. Sometimes, one looks good enough to try to see if we can detect its effects on the functioning of cells we grow in cultures. We write up our findings. Other labs try to replicate our results. We try to replicate theirs. Last year there were 800 scientific papers published on schizophrenia genetics. It’s a lot of work. But then, 500 million people in the world suffer from schizophrenia and we owe it to them, and to their parents, to keep on trying.