Mad Prime: Tag genetics

Cystic Fibrosis treatment -- sometimes the specific variant matters

Madeleine Ball — Wed, 08 Feb 2012 14:44:00 -0500

Great news for cystic fibrosis research! And also an interesting demonstration of how two different variants causing the same disease can end up having different clinical consquences.

Drug bests cystic-fibrosis mutation (Nature News)

In GET-Evidence (our genome/variant review system) we currently score each variant's clinical effect individually. This tends to cause a lot of redundant labor -- most of these scores (severity, treatability, and penetrance) are the same for all variants causing that disease. As we expand the system, one suggestion has been to add disease pages and have all variants refer to those pages for clinical importance scores. Generally this sounds like a good idea.

This case is an exception that shows how variant-level information will still be important. In this case CFTR-G551D is found to respond much more effectively to this treatment than CFTR-F508Del. (The latter is the one causing disease in most patients, unfortunately.) This means CFTR-G551D would score more highly on "treatability" than CFTR-F508Del. Even though both cause the same serious disease, the treatability of that disease depends on the particular genetic variant.

The decline of the microarray

Madeleine Ball — Wed, 25 Jan 2012 13:47:00 -0500

Via David Nusinow on Google+:

When can we expect the last damn microarray paper? (http://jermdemo.blogspot.com/2012/01/when-can-we-expect-last-damn-microarray.html)

2016 for the last microarray paper? Sounds a bit optimistic, but we can dream. Of course there will likely be niche uses for microarrays for a while yet (e.g. cheap linkage analysis in a pedigree, as in my last post...) -- the analysis here is just counting usage of the word in the title (presumably indicating the technology is a major focus of the paper).

It's a Boy!: or How I Learned to Stop Worrying and Love Direct to Consumer Genomics

Madeleine Ball — Sat, 21 Jan 2012 14:32:00 -0500

This is going to be a long, complicated post so I'll get the simple personal news out of the way first: It's a boy!

Normally such news would be interesting, but neither good nor bad. In my case, however, I was really hoping for a girl – for some concrete reasons of family history. I have two uncles (my mother's brothers) with mental retardation, likely caused by an unknown mutation. If that mutation is X-linked, there's a chance I'm a carrier for that X-linked mutation. If I have a boy, there's a chance I could pass it on to him.

Assuming it is X-linked, what are my chances of being a carrier? My mother had a 50% chance of inheriting it and I had a 50% chance of inheriting one from her – a total of 25% chance (50% of 50%). However I have something in my favor that makes this chance lower: I have two brothers who are clearly unaffected. Applying this new information to a Bayesian calculation, my mother's chance of being a carrier becomes 20% and my own becomes 10%.

I would have a 5% chance of passing it on to a male child (50% of 10%). Of course, whatever they have might not be X-linked (see footnote). I estimate the chances their issue is X-linked is around 50%, which would make the risk 2-3% instead of 5%. This is small but not trivial – it's like the risk a 44-year-old woman would have for Down Syndrome in her child. It is standard to offer such an expectant mother the option to test her fetus and choose to abort it (a common decision, if controversial). I have no such option, we can't test for a mutation if we don't know what it is.

And so, when I told my mother the sex, she decided to try to have her brothers diagnosed ASAP. It's not something she wanted to do, it would be difficult – it would require discussion with their caretakers, arrangements with medical providers, and it would disrupt their routine.

I was thinking hard about what I could do myself. My mind challenged with this new stressful information, I thought of something I could have done before now but hadn't occurred to me – I could use the genetic data from my family members to determine how much “at risk” X chromosome DNA I carry.

A map of the X inheritance in my family is below (apologies to colorblind people, I believe you can read “magenta, red, and cyan” as “blue, yellow/black, and gray”):

My grandmother's X's are represented as magenta (safe) and red (mutation bearing) – my aunt and mother are represented as inheriting a red/magenta mixture from her (we don't know which they got) and a cyan X from my grandfather. My sister and I inherited (1) an X from mom that is a mixture of my grandfather's cyan X and the unknown red/magenta from my grandmother, and (2) an X from our father. My brothers inherited a single X from my mother's parents – and here I've colored it as only magenta & cyan, because any parts they got from my grandmother are proven to be “safe” by the fact that they are not affected.

Such a diagram would lead you to think that inheritance blends the DNA up pretty well each generation, but it really doesn't – in reality inheritance is extremely blocky. 23andme provides diagrams of these blocks using the “Family Inheritance” and “Family Inheritance: Advanced” tools. Because my grandfather and brothers and I have all done 23andme, I can take the blocks I inherited from my grandmother and compare them to what my brothers inherited. If either of them inherited the same regions, then I knew those parts were confirmed “safe”. I only have to worry about regions I inherited from my grandmother and both brothers did not inherit.

Based on 23andme, I have the following maps for the X chromosomes of myself and my brothers. What I did was perform a comparison between each of us and our grandfather (all of whom have 23andme data); any regions matching my grandfather were marked as inherited from him, any remaining regions marked as from my grandmother.

As you can see, it looks like I'm in the clear! Amazingly, brother 2 (his name is Andrew) managed to “test” a whopping 95% of the X our mother inherited from our grandmother! Thank you Andrew, excellent work! While there's 5% from our grandmother that neither brother has, I don't have it either so it's not an issue for me. Our sister does carry that segment -- an eventual diagnosis of our uncles would be useful for her, but there is no pressure to do it quickly.

Because diagnosis of my uncles was never certain, it's possible this has saved me years of uncertainty as I watched the child for signs of developmental delay. I'm extremely relieved -- and I'm really happy the "direct to consumer" genetic testing industry exists, allowing me to do this analysis. The result could easily have fallen in the other direction: if I found a larger-than-expected untested fraction from our grandmother then my risk estimates would have increased rather than disappeared. I am lucky it turned out the way it did. It's a happy ending for me, and now I can return to the mundane common concerns of pregnancy and children.

Addendum: I was asked how much of the second diagram required me processing raw data myself from 23andme. Very little, in fact -- all I needed was the comparison using 23andme's "Family Inheritance: Advanced" tool. (I could theoretically have done this myself if I got the raw data from my grandfather and brothers, but they'd have to download the files and send them to me, and I doubt I'd do a better job). Below is the screenshot of the X chromosome when I compare my grandfather with (top) myself, (middle) brother 1, and (bottom) brother 2. The second diagram was drawn using this as a template.

Footnote: X-linked disorder was the major concern here. The other major possibility would be an “autosomal recessive” disorder, but the chances I would have an affected child even if I carried such a gene would be extremely low – Chris would also have to be a carrier for a mutation in the same gene (presumably an extremely rare event).

Other possibilities I could worry about carrying include Fragile X and balanced translocations. Fragile X is a string of trinucleotide repeats that can expand to a disease-causing length when passed from mother to child – I suppose it's theoretically possible one or both of my brothers and I inherited “premutation” versions and that mine could expand into “mutation” when I pass it to a child. It's also theoretically possible that my grandmother had a “balanced translocation” where a disease is caused if you only get one or the other, but not if you get both – in such a situation it's possible (but very unlikely) that my mother got both, somehow only passed zero or two to each of her children, and passed the same balanced set to me. I think both of these are very unlikely, but I am being tested for them and will get results in two weeks. I don't worry about the outcome of these tests, the scenario that made me so stressed was the inability to know whether I was carrying or passing on a mutation – to me, being able to test means everything.

8-bit model organisms

Madeleine Ball — Wed, 07 Jul 2010 23:12:00 -0400

I don't know how I found time to do this. It didn't take that long. I was supposed to do a 4th-of-July patriotic theme but got distracted.

Thesis defense

Madeleine Ball — Wed, 21 Apr 2010 11:30:00 -0400

I defended last Thursday. People claim I gave a good talk. My sister (who is pretty nonscientific) claims she understood large segments of it!

Chris recorded the talk with audio and video. It was hard to read the slides on the original video, so he's created a video using the audio only and pics of the slides themselves.

Madeleine Ball's thesis defense from Madeleine Ball on Vimeo.

I should watch it myself, but dislike the sound of my own voice... I think I sound like a chipmunk. Honestly, in my head my voice is much deeper than that.

Anyway, if you take the time to watch, I hope you enjoy it! The subjects are epigenetics, DNA methylation, and a little on clinical analysis of whole genomes (related to the Personal Genome Project).

5% Irish Cream

Madeleine Ball — Thu, 17 Dec 2009 16:00:00 -0500

Hah! It looks like there's a protocol for DNA hybridization in blots using 5% Irish cream liquor! I kid you not. Here's a sample quote from the methods of Yamamoto et al. 1993:

Hybridization was carried out overnight at 65°C in a solution containing 6 × SSC, 5% Irish cream liqueur (Original Irish Cream, R & A Bailey's), 20 mM Na₂HPO₄, 20 μg/ml heat-denatured salmon sperm DNA, and 2 μCi/ml of the ³²P-labelled probe.

The original source appears to be Elbrecht, A. 1987, "Lab Hints: Irish Cream Liqueur as a Blocking Agent for DNA Dot Blots." BM Biochemica, 4:12-13. BM = Boehringer Mannheim, it appears to be a newsletter. It's too obscure for my cursory searching to turn up a copy of the original, I wonder what the motivation was! Maybe this way you can order liquor using grant money? This idea has a lot of potential...!

ExploreTree & pretty flowers

Madeleine Ball — Tue, 08 Sep 2009 23:13:00 -0400

The New York Times has a nice article on flower evolution today.

If you enjoy looking at evolutionary trees to see how closely related different living things are, you might enjoy playing with ExploreTree. I've added features that make it a lot more fun: the zooming in and out is animated, you can search for an organism and follow a path. Plus now, with a little help from Chris, it runs on a webpage (feel free to show it to friends & family). Give it time to load, though.

Here is a snapshot of the location illustrated in the NYTimes article:

I've put off posting about the program for a while since I kept hoping to improve it a little more, but here it is. It was written in processing, you can get the code if you'd like to play with it here (or improve it!) on github.

Genetics Luau!

Madeleine Ball — Thu, 30 Jul 2009 02:24:00 -0400

Another social hour poster illustration:

Harvard Genetics Retreat 2009

Madeleine Ball — Thu, 28 May 2009 02:55:00 -0400

Was kind of sad this year having a "retreat" in a slightly different building in the same city. I amused myself by constructing a genetics "buzzword bingo" list.

Slide showing a signaling pathway with >= 15 proteins named
Messing with this gene causes cancer
Anything involving stem cells
Microarray data
High throughput / deep sequencing
RNAi
Animal model vaguely resembling a human disease
GWAS
Messing with this gene makes this tissue/organ look funny
Epigenetics
Mass spec data
FACS
Evolution
A photo that makes you lose your appetite
Apoptosis is mentioned
HAIRBALL (aka. "interaction network")
Anything related to sex (eg. chromosomes)
RNA splicing
Bacteria
Yeast
Plant
Worm
Fly
Fish
Mammal

Although silly, I found this actually helped keep me paying attention to talks. I applied it to the last session and almost got a BINGO, but Norbert Perrimon's signaling pathway slide only had 13 proteins. So close!

Color Blindness

Madeleine Ball — Mon, 04 Aug 2008 01:16:00 -0400

For some time now, I've been wanting to write about red-green color blindness, a dramatic perceptual difference with an interesting genetic and evolutionary story. This first post will mostly be an introduction to the topic. If you are color blind: I always feel guilty when I speak of this as a deficiency, or when I emphasize how profound the differences seem to the rest of us. I hope it doesn't bother you. I always wish I could pee standing up so... there.

Daylight vision in humans is mediated by the opsin proteins, which transmit signals that activate nerves when they are hit with light. Humans have three different opsins with different sensitivities to the colors of the spectrum -- it is the different color sensitivities that allow us to see color. You can call these the "blue", "green" and "red" opsins.

A normalized diagram of the sensitivities of opsins to different wavelengths of light.
"S" = "short wavelength", is the "blue" opsin.
"M" = "medium wavelength", is the "green" opsin.
"L" = "long wavelength", is the "red" opsin.

In its severe form, red-green color blindness occurs when a man is missing the "green" or "red" opsin - these conditions are respectively known as deuteranopia (1% of all males) and protanopia (another 1% of males). They are fairly similar in effect: a total loss of ability to distinguish hues in the green to red range. There are many less severe forms of color blindness -- 6% of males -- but that's a later post.

I say "males" because color blindness is almost always seen in men. This is because the "red" and "green" opsin genes are located on the X chromosome, which men have only one copy of. Women have two X chromosomes; even if one has inherited a deletion mutation, the other can serve as a back-up. For a woman to be color blind, both X's would have to carry the same mutation, which is much less likely to occur. (e.g. 1% * 1% = 0.01%)

I'll end this post by showing you what color blindness looks like. Vischeck is a service available online that simulates how images look to a color blind person. To a color blind individual the simulation and original images should look identical (or nearly so - computer monitors vary, so this cannot be perfect). If you're curious about the algorithm, the program is based on this paper.

Deuteranopia	Original	Protanopia

All colors in the red to green range -- green, yellow, orange, red -- are simulated here as yellow. As you can see, deuteranopia and protanopia are almost identical - the main difference is that red looks darker to the protanope (look closely at the picture of cars). Also interesting to note: the butterfly picture demonstrates how purple looks like blue to the color blind individual.

Credits: Opsin sensitivity diagram adapted from Wikipedia diagram, credit goes to User:Vanessaezekowitz and from the screenshot for Wavelength 1.3. Photos taken from flickr users Marshall Flickman, Teo, and Oneras under CC and CC-by-SA licenses.