Analyzing my DNA | Hacker News Reader

Analyzing my DNA

blog.alexmaccaw.com

108 pointsrichardburton42 comments

>"What's interesting is that the older the species, the more the chromosomes, or at least this is the trend. "

This isn't true: there's no such trend (Fruit flies have 8 chromosomes), and the sentence belies an important misunderstanding of taxonomy. Also, plants tend to have high chromosome counts because of [polyploidy](http://en.wikipedia.org/wiki/Polyploid).

Additionally, the author seems to be latching onto some odd victorian-era esque ideas about biology here. There are several fern species which are younger than humans, and several salamander species, and thousands of other extant species. Modern fern species aren't "older" than humans simply because they had distant cousins in the fossil record who were, on the face, morphologically similar.

Evolution is a bush, not a ladder, and it doesn't make sense to say that any one of the end nodes (extant species) is "older" than any other unless you are talking about the very fuzzy barrier of speciation at which the majority of the pre-species' population could not breed with the population it was diverging from.

For anyone interested, 23andme is also starting to sequence exomes (the important parts of the protein coding regions) for $999, which can potentially provide much more interesting information than simply looking at SNPs. But from the website, it looks like they are still in closed beta: https://www.23andme.com/exome/

However, there are several blog posts describing the types of results and analysis the exome enables. For example, http://jchoigt.wordpress.com/2012/07/02/a-first-look-at-my-e... and http://blog.goldenhelix.com/?p=1282 but more can be found on Google.

I thought they started this in late 2011 and stopped the program a few months later. Has anyone been part of the program since then?

Nope. It was a Pilot and they are not sure about doing more exomes.

From my last chat with Brian Naughton (their lead informatics guy) about this, it sounds like they are planning on doing more sequencing in the future. But it could be whole genome and it may be geared more towards research (your selected based on your phenotype) than open to any customer.

One of the most interesting things I've learned about human genetics by joining a local "journal club" including several researchers who are members of the Behavior Genetics Association is just how little is certain about genetic influences on any human trait of interest. The review article Johnson, W. (2010). Understanding the Genetics of Intelligence: Can Height Help? Can Corn Oil?. Current Directions in Psychological Science, 19(3), 177-182

http://apsychoserver.psych.arizona.edu/JJBAReprints/PSYC621/...

looks at some famous genetic experiments to show how little is explained by gene frequencies even in thoroughly studied populations defined by artificial selection.

"Together, however, the developmental natures of GCA and height, the likely influences of gene-environment correlations and interactions on their developmental processes, and the potential for genetic background and environmental circumstances to release previously unexpressed genetic variation suggest that very different combinations of genes may produce identical IQs or heights or levels of any other psychological trait. And the same genes may produce very different IQs and heights against different genetic backgrounds and in different environmental circumstances. This would be especially the case if height and GCA and other psychological traits are only single facets of multifaceted traits actually under more systematic genetic regulation, such as overall body size and balance between processing capacity and stimulus reactivity. Genetic influences on individual differences in psychological characteristics are real and important but are unlikely to be straightforward and deterministic. We will understand them best through investigation of their manifestation in biological and social developmental processes."

It should be called Y-chromosomal Noah, since he and his sons would all trace through one mans genetics. And well there was genetics from four different women on the Arc... So the Eve makes sense.

"certain European populations have a mutation that causes lactase to be produced throughout their lives". Lactase persistence is interesting. There has been selection pressure for persistence in Northern European (eg: about 98% of Irish people have lactase persistence) populations but not in Southern Europe (under 40% in Italy) and one might even say "almost selection against" in Asia (under 10% in China). Apparently, the gene first evolved in the Near East, possibly tied to first domestication of the auroch and then spread radially from there into Europe. Convergent evolution, ie: lactase persistence using different genes, also came about in the Bantu population in Sub-saharan Africa which enabled this population to spread further south and overtake San populations (bushmen). Perhaps this is the same selection pressure in action.

Well, that's a bit late, but if you want an easy tool to have extra information about your 23andMe analysis, you can also use this : http://my.promethease.com

More info about what it can do here : http://www.snpedia.com/index.php/Promethease

SNPedia can also be of interest to anyone here, it is a wiki centralizing information about all known human SNPs (including literature references)

Disclaimer: One of the person running this is my colleague.

Are insurance companies doing this yet? I imagine a Gattaca-like future when I hear of this kind of stuff. Not that I fear it but I don't trust any company or government to do the right thing with the information...

I am curious how accurate it all turns out to be (it is noted that percentages play a role). I have only basic understanding of the math and biology so I am probably making tons of assumptions but from the post:

"One way of investigating this, is by studying identical twins. Since they have the exact same DNA, any differences between the two can be attributed to environmental factors."

I have identical twin sisters (from a fraternal twin father even, born on the same birthday...the odds). To me they look nothing alike but to outsiders sure maybe.

The points I am curious about:

1a. Doesn't DNA mutate? Even if twins start with identical DNA it could mutate from any number of environmental variables couldn't it?

1b. Couldn't it only mutate in just one, making them have different DNA in the end?

Like I said basic understanding. Does 23andme give a 'layman' introduction to any of this stuff? One of the other comments says it makes you read some stuff before getting certain results but is it the kind of stuff that 'normal' people can understand without having to read another book of references?

Sorry about the long post; stopping now.

>1a. Doesn't DNA mutate? Even if twins start with identical DNA it could mutate from any number of environmental variables couldn't it?

Human DNA does not mutate that often (If I recall correctly, the average human can expect to see ~230 mutations in their nuclear DNA in their lifetime, this includes that parts of DNA that we believe do nothing) What is far more common is related to epigenetics. Attached to DNA are various promoters and demoters that control how much a given gene is expressed. These can and do change based on envirement.

What is also far more common is a mutation in the mitochondrial DNA. However, the only thing mitochondria does for us is to help break down sugars, so mutations in mtDNA are not that significant.

So essentially even when it does mutate the chances that something important is overwritten is fairly low.

Does anyone know of an online resource that explains this kind of stuff in easy to digest format? Something like a ELI5 for DNA/Cells.

MITx is offering a bio course [1]

[1]https://www.edx.org/courses/MITx/7.00x/2013_Spring/about

edit: After re-reading I understand this is really not what you are looking fore, but i'm going to leave it here anyway

Always wanted to try that, but with $80 shipping costs for Germany that's just a little bit too much.

Yeah, it's a bit hefty to ship it outside US/Canada. Remember the price includes shipping it to you and shipping it back to 23andMe. DHL will also collect the package from your door (at no extra cost) when you're finished.

If you know a friend/relative (or two) who is also interested then you can order more than one kit and save a little on shipping.

I decided to pay the $80 to get it shipped to Ireland. I sent off my saliva sample a couple of weeks and I'm currently awaiting the results.

I feel you. I ordered the kit while I was in the US. I think of getting two kits for my parents in Germany though.

23AndMe doesn't tell you much about your DNA actually, they can only tell you some things about already known variations. If you really wanna know everything about your DNA, you should look at whole genome sequencing (WGS). WGS is usually done by companies like Illumina and analysis can be done by a company like Bina Technologies. The whole field is getting more and more data and analytics driven so there's lots of opportunities for software engineers and data scientists. Check out http://www.binatechnologies.com/vision to get a better idea of the field.

> A single change in a base pair can dramatically alter the protein created, and easily account for the differences between our two species [human and chimpanzee that is].

Not to be picky, but I seriously doubt that. ;)

While he's seriously incorrect, it is however worth remembering that a caterpillar's DNA bases don't change at all when it becomes a butterfly. It's all about the degree those proteins get expressed, in what ratios, and (crucially) when that makes an organism. Sort of like if I give you the ingredients to a recipe:

bread cheese ham egg

You could scramble the egg, put it with the ham and cheese and put it all between bread and have a sandwich. Or you could coat the bread with the egg and throw into a frying pan and make a Monte Cristo. Both sandwiches, but pretty different.

As computer types we want to believe that DNA sequence is an analog to computer code. It's not a perfect analogy. There is the currently very hot field of research into epigenetics which is looking into the chemical decorations on DNA that make parts active and inactive at certain points in time. These are heritable, but unlike the sequence, are routinely and intentionally altered by cells in your body (often in response to environmental stimuli). So the environment can make temporary changes that get passed onto offspring, potentially affecting early stages of development. The thing is, development is crucial and not all processes are reversible (e.g. the lens in the eye only gets made once). So what long term effect do these changes have? Nobody has the foggiest idea yet.

Great information. You increased my knowledge and my appetite.

I'd say that the computer code analogy is how most people understand DNA: a set of constants with values. This is how DNA is generally portrayed.

23andme also has upgraded their website but still in beta mode. You guys can opt in for it here https://www.23andme.com/you/beta_optin/

If you want to dig into the million SNP raw data from 23andMe any further than what is presented on the website, Jeff Hammerbacher and Konrad Karczewsk taught a Skillshare on doing your own personal genome analysis with Python and various other tools.

Slides here: http://www.slideshare.net/jhammerb/20130206skillshare

Github repo here: https://github.com/hammer/personal-genome-analysis

I'm awaiting my 23andme results right now. I'm very excited for my results and this post has inspired me to learn more about the science and research that has enabled us to learn so much about ourselves from our genes.

My friend recently got their results back and I was really impressed with how well 23andMe presents the data to you. They do a really good job explaining what the data means. In fact, in some sections, 23andme will only tell you your results after reading through the informational material they provide.

Hmm, I imagine that this is a complete scam. Or am I too skeptic? Where is the proof that this is 100% legit and not some mumbo-jumbo psyche stuff?

Definitely too skeptical. Every piece of research that 23andme uses to interpret your genetic data is backed by a wealth of citations and studies. Each "discovery" is rated on a confidence scale from 1 to 4, with experimental research being a 1 and undisputed, established research being a 4.

Nothing's stopping you from downloading your own genetic data as a text file and performing the same types of analysis yourself.

Also, Occam's razor.

For $99 you can get DNA analyzing; that is a cheapest price you can find anywhere imo. However, their service is extremely slow. You are looking at about 6-8 weeks until you get to see your result. I guess they are trying to offset the expenses. Nevertheless, it's a great service to find out your roots and health problems.

That doesn't even seem extremely slow to me. My gram buys crap from china on tv that takes 6-8 weeks. Maybe it's because I know little to none about the subject, but how long should it really take?

Ideally, it would be about a week of processing time (probably a little less). But, I assume they have a limited number of machines and a backlog. And they probably batch everything to make it as cost-effective as possible. At $99 they are running very close to cost.

I bought the deal when it was on sale for $49 and I remember it being faster than that, however.

The turn around rate is 1-2 weeks before they received venture capital and lowered the price. Although, I believe that $49 deal came with yearly subscription? And you have to pay a bit extra to see all the cool new stuff.

I recently did this also, heres the process in pictures: http://deanjerkovich.com/2013/01/20/getting-my-genome-sequen...

please excuse the technical inaccuracies - this was prior to me going on an adventure of learning exactly what this involved.

Having studied the techniques used to analyze DNA, I wish 23andme would send you your raw data. Nevertheless still awesome and crazy to see your own DNA being analyzed.

You have access to your raw data through 23andme. At least I do. It's a text file (sitting in a Truecrypt partition, in my case).

My apologies. 23andme does provide the raw data from the genotyping. What I think I meant to say was I wish they could do sequencing and give that data. Obviously I understand it's prohibitively costly though.

"My DNA is 2.9% neanderthal, which puts me in the top 91% of 23andme's members (not sure what conclusions to draw from that!)."

Checkmate, evolution deniers.

When has science of fact ever got in the way of someone who denies evolution?

These are the kind of people who put saddles on dinosaurs and have "Creation Museums" (http://creationmuseum.org/)

Funny wording. Wonder if he would draw different conclusions if he were told he was in the bottom 10% (equally possible, disregarding margin of error).

"Studies have shown that identical twins who grow up in separate families have an IQ correlation of 0.74, while adoptive siblings have no more similar IQ than strangers."

First off the bat, such studies are obviously social science, not science.

Secondly - there is this idea that one can boil entire brains down to one number like a CRC or checksum - the IQ number. Then you can rank them in order I suppose. It is obviously a ludicrous endeavor on reflection. It's like the Douglas Adams joke that the answer to life, the universe and everything is 42. When science actually makes progress on the brain, I'm sure biologists of the future will look on IQ like we look on phrenology.

Thirdly, these social science studies of twins mentioned were done by Thomas J. Bouchard, Jr. He's someone who writes op-eds for the Wall Street Journal - I guess Nature and Science are too full and his work crowded into there.

The second edition of the Mismeasure of Man by Stephen Jay Gould is a good book on this topic.

There was a hysterical conservative reaction to Gould's book just as there was a hysterical progressive reaction to the Bell Curve. Which shows this is really a political debate, not a scientific one. This is a political debate going back about 10,000 years, really. I am skeptical of any social scientific study that proposes it has found all the answers.

>Secondly - there is this idea that one can boil entire brains down to one number like a CRC or checksum - the IQ number. Then you can rank them in order I suppose. It is obviously a ludicrous endeavor on reflection.

Nothing of that diatribe is reflected in the statistic you just quoted. Pointing out a correlation between IQ doesn't necessarily mean IQ is some perfect heuristic to weed out the proles.

"'Studies have shown that identical twins who grow up in separate families have an IQ correlation of 0.74, while adoptive siblings have no more similar IQ than strangers.' ... First off the bat, such studies are obviously social science, not science."

This is an important and legitimate way to estimate broad-sense heritability. It won't control for environmental effects within the womb, or cultural effects, but it's pretty strong science absent those criticisms.

> such studies are obviously social science, not science.

Hey man, you can believe whatever makes you feel better.

The Mismeasure of Man is notoriously awful trash. It's amazing that anyone would recommend it as a reference or guide to the subject when its poor quality is so widely recognized by actual experts in the field.

Nobody believes that IQ expresses everything about our mental functioning. It is instead a useful and reliable way of measuring important aspects of cognitive ability.

Thomas J. Bouchard, Jr.

I know Thomas J. Bouchard, Jr., and he has been a participant in the journal club I mentioned in my top-level comment in this thread. He is a serious researcher on human behavior genetics and he has had publications in Science and other leading journals of peer-reviewed scientific research. I by no means claim that he would endorse all of my opinions about human behavior genetics, nor would I endorse all of his, but I will endorse him as a truth-seeker and straight shooter who attempts to take his opinions where the facts lead him, as he best understands the facts.

A lot of the rest of the tone of your reply is just setting up tribal affiliations and name-calling. But on the substance of what you wrote, perhaps you and I could agree in endorsing a review article by Eric Turkheimer (a colleague and occasional co-author of Bouchard's, and current president of the Behavior Genetics Association):

Turkheimer, E. (2012). Genome wide association studies of behavior are social science. In K. S. Plaisance & T.A.C. Reydon (Eds.) Philosophy of Behavioral Biology (pp. 43-64). New York, NY: Springer.

http://people.virginia.edu/~ent3c/papers2/Turkheimer%20GWAS%...

"If the history of empirical psychology has taught researchers anything, it is that correlations between causally distant variables cannot be counted on to lead to coherent etiological models."

Reply on news.ycombinator.com