Monday
Nov282016

The Data Science Behind Cyber Monday

As you awaken from your Thanksgiving food-coma and return to the work-week grind, don’t forget: today is Cyber Monday!  The holiday-shopping season officially began last week on Black Friday, and today retailers are expected to reign in similar profits.  Forbes magazine predicts that retail sales will exceed $3 billion on both Black Friday and Cyber Monday [1].  

And in the spirit of giving this holiday season, shoppers will be handing over enormous amounts of data to every site they visit.  Behind the scenes at online retailers like Amazon are teams of data scientists tasked with pricing merchandise, targeting advertising, and predicting consumer behavior [2].  Ever notice that when you visit an online retailer and browse for an item without buying it, that item will show up in advertisements on other webpages for the next few days?  That’s a machine-learning algorithm at work, designed by data scientists.  As you browse the web, you leave a trail that can be picked up by advertisers.  So when you compare different cell-phones on Amazon, odds are you’ll start seeing ads for mobile devices in your Facebook newsfeed and in banner-ads while you’re reading the New York Times. With more data, these algorithms “learn” faster, so they actually become more efficient during the holiday-shopping rush [3]. The algorithms consider everything from the items you browsed and the stores you visited to the movement of your mouse and the number of clicks you made on a particular page.  On sites like Amazon, they’ll also consider whether you’re a Prime Member and items you’ve previously purchased.  

Thanks to Big Data and advances in machine learning, advertisers can push their products to targeted demographics like never before.  Still, this type of advertising does have some drawbacks.  The large amounts of data that companies accumulate about their shoppers could come with some security risks.  Moreover, last month Facebook came under-fire for offering the ability for advertisers in housing, employment, and credit, to target by “ethnic affinity” [4].  And in 2012, Target’s algorithms accidently exposed a teen girl’s pregnancy to her father [5].

Despite these drawbacks, advances in machine learning and data science and paving the way for new discoveries in other fields.  Similar algorithms power Siri and the Amazon Echo.  Machine learning and Big Data are revolutionizing the way we understand genetics, disease, and the human brain.  Like any new technology, machine-learning algorithms should be designed with their context in mind and implemented with care.  So as you shop this Cyber Monday, you can start by crossing one gift off your list: the gift of data that you’ll be giving to all of the retailers you visit.
-TM
[1] http://www.forbes.com/sites/shephyken/2016/11/19/2016-black-friday-cyber-monday-holiday-insights-and-predictions/#785131c5243f
[2] https://www.internetretailer.com/commentary/2016/11/12/science-behind-black-friday-and-cyber-monday-pricing
[3] http://everything-pr.com/big-data-black-friday/54064/
[4] http://www.forbes.com/sites/kathleenchaykowski/2016/11/11/facebook-to-ban-ethnic-affinity-targeting-for-housing-employment-credit-related-ads/#2980ba0ca747
[5] http://www.forbes.com/sites/kashmirhill/2012/02/16/how-target-figured-out-a-teen-girl-was-pregnant-before-her-father-did/
 

 

Wednesday
Nov232016

The science behind the food coma

Tomorrow is Thanksgiving, a day of family, football, and feasting! Following a large meal, many people experience what is known colloquially as a “food coma”.

Recently, scientists from the Scripps Research Institute, Florida Atlantic University, and Bowling Green State University may have found a reason for the phenomenon of postprandial sleep.  William Ja and colleagues used Drosophila (fruit flies) as a model to investigate the effects of eating on sleep. They found that after a meal, flies increased sleep for a short period before returning to a normal state of alertness. Flies that ate more also slept more. Protein, salt, and the amount eaten increased sleep, but sugar had no effect.

Researchers also used genetic tools to turn on and off neurons in the fly brain and identified a number of brain circuits that play a role in controlling post-meal sleepiness. Some of these respond specifically to protein consumption, while others are sensitive to the fruit fly’s circadian rhythms.

While this study was in fruit flies, there are some parallels and connections to mammals. Researchers speculate that post-meal sleep is important, perhaps for boosting digestion or helping animals form memories about a food source.

I wish you and your family a Happy Thanksgiving! Hopefully the “food coma” won’t hit too hard. 

CJW

References:

Scripps Research Institute. “Scientists Find Surprising Answers to ‘Food Coma’ Conundrum.” NeuroscienceNews. NeuroscienceNews, 22 November 2016.
http://neurosciencenews.com/food-coma-neuroscience-5578/

“Postprandial sleep mechanics in Drosophila” by Keith R Murphy, Sonali A Deshpande, Maria E Yurgel, James P Quinn, Jennifer L Weissbach, Alex C Keene, Ken Dawson-Scully, Robert Huber, Seth M Tomchik, and William W Ja in eLife. Published online November 22 2016 doi:10.7554/eLife.19334

 

Monday
Nov142016

A Primer on Hormonal Hunting on World Diabetes Day! 

 

As a diabetic, I dread the feeling of hypoglycemia. It happens when I spend hours in the lab and forget to eat, when I overestimate my appetite and give too much insulin or when I exercise without making sure to give my body some extra sugar. Non-diabetic colleagues of mine claim to know the feeling of just craving sweets, feeling week and being unable to focus. I don’t know if they fully understand how it feels though when you break into cold sweat, you feel like you can’t hold yourself upright and certainly can’t keep your thoughts organized. It’s awful. So awful, evolution decided that it would be a great way to weaken prey and make it easier to hunt!
Snails aren’t exactly fast enough to catch fish under normal circumstances, but cone snails use venom to disorient small schools of fish and use their extended mouth like a net to catch their prey. Conus geographus and Conus tulipa, two cone snail species, have been found to use a very remarkable strategy to disorient their prey – they target their energy metabolism to induce hypoglycemic shock. The component of their toxin that is responsible for that reaction in fish is a modified version of insulin. Interestingly, the peptide is much more similar to fish insulin than the mollusk’s own, yet bears the typical posttranslational modification signature of the snail’s usual toxins. 
While I certainly empathize with the poor fish who die in this dreadful manner, I would very much die without insulin and am therefore very thankful to the founders of Genentech, who enabled humanized insulin production through their recombinant DNA technology which is now used by Eli Lilly to keep millions of diabetics healthy and happy. For humans insulin isn’t exactly a great weapon: A review from 2009 stated that only 66 cases of homicide by insulin have been reported, in which 11 needed an additional weapon. As I said… It’s slow and painful, but definitely gives you enough time for a 911 call. 
MGS
1) PNAS 2015 112(6), 1743-8
2) Drug Test Anal 2009 1(4), 162-76
Wednesday
Oct262016

Happy Halloween!

To get into the Halloween spirit this weekend, my roommates and I carved pumpkins. We had a lot of fun as we carefully cut out our designs, while trying to preserve our fingers! However as you know, before you get to the fun of carving there is a lot of prep work to be done. Our pumpkins produced A LOT of pumpkin “guts”. We didn’t want to waste all of this, so we decided to roast the seeds. YUM, this definitely was a good decision!  I was curious as to the nutritional value of pumpkin seeds and I learned they are a great source of protein, fiber, zinc, etc. In fact, a research study was published this summer by XJ Zhao et al. suggesting pumpkin seed oil (PSO) can alleviate certain types of cellular damage caused by a high-fat diet in rats1. This study divided rats into three groups: 1) controls fed a normal diet (n=20), 2) a high-fat diet group (n=20), and 3) a high-fat diet group with PSO intervention (n=20). Liver tissue samples were examined by histology and quantitative real-time PCR. The PSO intervention group showed decreased accumulation of fat deposits in the liver, as compared to the high-fat diet group1. Further, the PSO intervention group displayed normalized expression patterns of genes involved in lipid metabolism and inflammation, as compared to the high-fat diet group1. While it is unknown whether these results will translate to humans, consider roasting pumpkins seeds as you celebrate Halloween this year!

-TMG

1) Zhao et al. (2016) Intervention of pumpkin seed oil on metabolic disease revealed by metabonomics and transcript profile J. of Science of Food and Agriculture

Friday
Oct072016

Nobel Price for molecular machines

On Wednesday this week, the Nobel Prize in Chemistry 2016 was awarded to Jean-Pierre Sauvage, Sir J. Fraser Stoddart, and Bernard L. Feringa for the design and production of molecular machines. What are molecular machines? These tiny machines are a thousand times thinner than a strand of hair and made of linked molecules with movable parts. 

Since the mid 20th century, chemists have been attempting to produce molecular chains in which ring-shaped molecules were linked together. Normally, covalent bonds hold the atoms in molecules together. In these chains, chemists wanted to create mechanical bonds, where molecules were interlocked without directly interacting with each other. In 1983, Jean-Pierre Sauvage used a copper ion to create molecular chains with a yield of 42%. These molecular chains, called catanenes, were early type of non-biological molecular machine. In 1994, Jean-Pierre Sauvage’s group succeeded in producing a catenane in which one ring rotated around the other ring when energy was added.

The second major step was completed by Fraser Stoddart in 1991, when he developed a rotaxane, in which a molecular ring was threaded onto a thin molecular axle. An electron-poor ring was threaded around an electron-rich axle. The addition of heat could be used to control the movement of the ring along the axle. Molecular lifts, artificial muscle and a molecule-based computer chip have since been created using rotaxanes.

Bernard Feringa was the fist person to develop a molecular motor in 1999. The motor consists of two flat chemical structures joined by a double bond between two carbon atoms. Methyl groups attached to each rotor blade function as ratchets that keep the molecule to keep rotating in the same direction. Exposure to UV light pulses cause the rotor blades to move 180 degrees around the central double bond. His group has optimized the motor so that it now spins at 12 million revolutions per second. Using molecular motors, he has rotated a glass cylinder that is 10,000 times bigger than the motor and also designed a nanocar.

The Laureates have started a toolbox of chemical structures that can be used to build increasingly advanced creations, such as a molecular robot that can grasp and connect amino acids and intricate webs of molecular motors that wind long polymers.

In the 1830s, when the electric motor was at the same stage, scientists could display various spinning cranks and wheels, but had no idea that the electric motor could like to instruments like washing machines, fans, and food processors. It’s interesting to imagine what the future could hold for molecular machines!

-CW