Thursday 23 April 2020

Guilt by flush

Privacy is being constantly redefined as it gets ever easier to pick up information about us. Back in 1994, when the OJ Simpson murder trial was flooding the airwaves, there was a bit more focused over-coffee discussion in the Genetics Department. Not many people, in those distant days, would have known what DNA evidence was and how it was gathered; and even fewer would have been able to evaluate the statistical analysis on which a large part of the prosecution's case depended. The then Professor of Genetics suggested that we should all have our cheeks swabbed and the DNA deposited in a secure database. That way, the police would have a much easier task collaring offenders . . . and the rest of us (all innocent of any crime) would have nothing to fear; and so nothing to lose by rowing in behind this scheme. There are many things wrong with this ostensibly beneficial implementation of Big Data. My first worry (which might have been several hours after the scheme was proposed, because I R slow) was about the secure database.  I found it hard to believe then that the Gardai of 1994 would have been able to establish a truly secure database. Twenty years later, my misgivings were vindicated when it transpired that individual gardai were using the secure police database to not mind their own business about various celebrities and the guard's own neighbours. The cited bloboprev was mainly about ubiquitous CCTV cameras and whether they could be used for gossip and tittle-tattle about the lawful but slightly sketchy actions of other citizens.

In 1994, we were in the ha'penny place with what we thought were Big Data, the human genome was more than a twinkle in Robert Sinsheimer's eye but still years from completion . . . and that was just one genome down [hey, thanks Bill and Tony], 7.5 billion to go. 23andMe, the direct-to-consumer genetic testing company, was founded in 2006 and now has thousands of DNA samples on its books each paid for by the punter. But the company is monetising the information by making data available to Big Pharma to find associations between mutations and disease. What's not to like about that? Nothing, I guess, except privacy and the laws of unexpected consequences. When you send your cheek-scrapings and $100 to one of these DNA sequencing companies you have to sign a long-and-long informed consent document. Which is far too long for most people to read with care and attention . . . and in any case there are sunk costs of $100 and on some level the decision has been taken about doing the test. So far 23and Me have resisted the blandishments of law enforcement agencies to access the data to find matches to rogue semen samples and crime scene blood spatter.

But your information is Out There in a world where every week there is another cyber-security breach with millions of files of person data travelling further than Best Intentions Inc. intended. And here's the thing, when you don't read the GDPR terms and conditions  attached to your DNA make sure that your children don't read them either; and the same goes for your brothers and sisters who also have half their genetic variants in common with you. And maybe your nieces, nephews and grandchildren have locus standi too. By unwittingly grassing yourself up as having a genetic predisposition for Condition X you are giving up your family as hostages to fortune. Quite apart from tying one of you to a rogue semen sample in the form of oh what a beautiful baby.

This all came bubbling up in my mind on reading a report about covid-19. Arragh Jaysus, Bob, enough with the Coronarama! Sorry lads, resistance is useless in these troubled days. One of the key problems with discovering the impact of covid-19 is working out how many people have been infected. John Ioannidis is now convinced that loadsa people in Santa Clara Co. CA are infected with the virus (and by implication other counties across the USA) which means that covid-19 is no more fatal than season 'flu [calcs tutorial]. One of his caveats is a possible informed consent bias in the Santa Clara data: not everyone would have agreed to have a cotton bud inserted in their throats beyond the gag-reflex.

What the IFLS link reports is a study based in MIT but involving 16 authors from 8 different institutions in 2 countries. They went to a local wastewater treatment plant before and after covid-19 and sampled the dulite shite dilute shute dilute shite. They used RT-qPCR to hunt out covid-like DNA and sequence them. The q in RT-qPCR is quantitative, so their methodology gets a titre / concentration of the virus in that community's sewage. They extrapolate to find, as in Santa Clara Co, that there is waaayyy more SARS-CoV2 out there in Massachusetts than the official confirmed cases would imply. Because The Man, at least in Ireland, doesn't have the will and/or the resources to sequence everyone it is fatuous to base mortality stats on tests limited to sick people admitted to hospital and their front-line carers.  So much, same hymn sheet.

But the elephant in the room of the MIT study is that pretty much anyone can pop over the fence of pretty much any waste-water treatment plant [there is security but it's not like Porton Down or Area 51] and find out who lives in the community. I'm not thinking so much about Natural Born Killers holed up after a spree, as people with a tendency to clinical depression, carriers of CFTR cystic fibrosis and generally people who want to mind their own business.

No comments:

Post a Comment