Saturday, April 30, 2016

Does Animal Research Count as Unethical?

The lecture about responsible and reliable research presented by LeBreque sparked a lot of great conversation on the blog about unethical research and how much has changed in the past 50 or so years. Studies like the Tuskegee Study of Untreated Syphilis in the Negro Male resulted in better regulation of  things like informed consent in studies involving human subjects in order to try to prevent experimental issues like this in the future. A comment on another blog post how the use of subjects, both human and non-human, is an evolving topic, and it made me wonder whether or not use of animals for scientific research will one day be considered unethical by the majority. I use mouse models in my research, and while I am not happy about it, I understand that they are being used for the greater good of advancing research, even if they are being used without their consent, and regardless of the fact that we are giving them diseases they most likely would not have contracted otherwise. 

I decided to look into some of the current concerns related to use of animal research, and found an article posted by the New England Anti-Vivisection Society (NEAVS). According to the article, 24 of the 30 Institutes and Centers that make up the NIH use animals in their research, and about 40% of NIH-funded grants and contracts involve animal research. We use animal models under the assumption that they recapitulate humans biologically, and while we know animal models can represent human diseases and responses well, in many cases the way animals respond to treatment does not mimic human responses. The article gives examples such as how "forcing dogs to inhale cigarette smoke did not show a link to lung cancer; Flosint, an arthritis medication, tested safe in monkeys but caused deaths in humans; and the recalled diet drug fen-phen caused no heart damage in animals, while it did in humans". More than 25 million animals are used yearly in the U.S. in all areas of research, testing, and education, and the article says that statistically, 92% of drugs tested in animal and approved for human trials fail. Outside of drug studies, I would be curious to see how much of the data obtained through use of animals successfully models outcomes or responses in humans.

The article points to how factors like stress, which is often experienced by animals in labs, negatively influences the reliability of animal research data. According to the article, stress influences heart rate, pulse, blood pressure, muscular activity, and hormone level, and can influence results observed in studies. Cost of using animals in research is another issue addressed by the article. Animal research is a multi-billion dollar industry, with interest from the animal importers, breeders, dealers, cage and equipment manufacturers, feed producers, and drug companies. Additionally, purchase and maintenance of animals is expensive. "Rats, mice, and birds are the main animals used, not because they are necessarily the best or most reliable, but because they are relatively inexpensive to buy, easy to manage and maintain, and because they are disposable without much public clamor or concern."

Another point made by this article was related to the development in the last 10 years of using transgenic models for medical research. While transgenic models are great for attempting to remedy the biological differences between humans and animal models, the NEAVS article pointed out that if transgenic models are needed to genetically improve existing animal models, perhaps animal models really aren't justified for research use, because they really don't have biological and/or clinical relevance.

I agree that animal models are not perfect for researching human conditions, but I also cannot think of a non-animal alternative that is a better solution. I also believe that there should be some sort of in vivo model used after in vitro studies before testing drugs or whatever else in humans. The article blames a lack of alternatives on the investigators, but I don't see them offering any alternatives either. I guess time will tell whether animal research is eventually considered unethical, but I don't think that will come until someone comes up with a good non-animal alternative that does not jeopardize progress made with testing human subjects. 

Friday, April 29, 2016

Non-parametric ANOVA: The Kruskal-Wallis test

Although we didn't have time to cover it in class, I wanted to briefly introduce the Kruskal-Wallis test, which is the non-parametric equivalent of a one-way ANOVA. I found this test while working on the BadStats assignment, where in I found a paper that tried to analyze survey data based on a 10-point scale. The paper totally butchered their analysis (trying to use parametric tests on non-parametric data), but I found that they should have used the Kruskal-Wallis test instead.

So, from a practical standpoint, I want to walk through how to use the Kruskal-Wallis test on Prism. Fortunately, Prism is a pretty fantastic software package that makes a lot of decisions for you when setting up this test. For this example, I made up a totally hypothetical experiment where we want to test how graduate student happiness varies between different years of students. Let's pretend we interviewed 20 students each from their 1st year, 3rd year, and 5th year of grad school and asked them to rate their happiness on a scale of 1 to 10, with 1 being "grad school has crushed my soul" and 10 being "I never want to leave grad school, this is amazing". Below are the survey results in a column table in Prism

Next, select Analyze --> Column Analysis -->One-way ANOVA (and nonparametric)

This data is not paired, so choose "No matching or pairing". And then below, do not assume a Gaussian distribution (this is nonparametric data and it totally doesn't follow a normal distribution), so choose the "No. Use nonparametric test" option (really, it's that easy to switch from parametric to nonparametric tests on Prism!). As you can see, Prism suggests the Kruskal-Wallis test, which is exactly what we're looking for.

If you want to make multiple comparisons, select that menu from the tabs at the top. In this scenario, we want to compare the mean ranks of each column to the others. You also have the option of making select comparisons or no multiple comparisons at all, depending on your experiment.

In the options menu, I chose to correct for multiple comparisons with the Dunn's test (which is the proper choice for planned multiple comparisons with the Kruskal-Wallis test). If you don't want to make corrections for multiple comparisons, choose the Fisher's LSD test  (third option down). I also chose to report the multiplicity-adjusted p-values since we are making multiple comparisons and don' want to p-hack the established alpha level of 0.05 for the whole experiment.

Here's the Kruskal-Wallis and multiple comparison results (also note that I graph the RANKS and not the raw survey data, as is proper for non-parametric data):
As you can see, these hypothetical grad students show significant variance in the ranks of their happiness levels between years, showing that this hypothetical grad school is a soul-crushing machine that increases its effects over time. Bless you, imaginary 5th years, I hope you escape soon.

Hopefully this was at least slightly helpful in demonstrating how to set up a Kruskal-Wallis test in Prism. Here's to hoping that scientists will actually use non-parametric tests on ordered and non-Gaussian data!

Non-parametric test flow chart

In class we discussed the importance of using non-parametric tests on ordered data (i.e. data from a subjective 1-to-10 scale). We talked about the variety of tests available, but I decided to make a handy flow chart for anyone who wants a quick reference of which test to chose.

As you see above, the first deciding factor is the number of factors. This weeds out your possibilities really quickly since you can only properly do one-factor non-parametric tests. For reasons that are really complicated, there simply isn't a way to do the equivalent of a two-way ANOVA for non-parametric data; signed-rank tests aren't made to produce enough information to have any meaningful analysis across multiple factors. A search through the internet shows that some people have attempted to develop multi-factor non-parametric tests, but statisticians have deeply contentious opinions on whether they work or not, so it's best just to avoid them. So, keep this in mind when designing studies that all non-parametric data should only be tested by one factor at a time.

Next, your test depends on the number of sample groups you want to compare. If you only have one group and want to compare to a set/expected value, then use the Wilcoxon signed rank test (non-parametric equivalent of a one-sample t-test). If you have two groups, you can use the Mann-Whitney U-test for unpaired samples (equivalent of unpaired t-test) or the Wilcoxon matched-pairs signed rank test for paired samples (equivalent of paired t-test). If you have three or more groups, there actually is a non-parametric equivalent of a one-way ANOVA! We didn't talk about it much in class, but it is called the Kruskal-Wallis test, which I will address in my next blog post. If you are making multiple planned comparisons, use the Dunn's test to correct for the multiple comparisons. If you are not making multiple comparisons, you can use the Fisher's LSD test.

Hope this flow chart helps my fellow visual learners out there!

Thursday, April 28, 2016

A Stock Solution; A Different Topping on Every Slice

When I was 14, my soccer coach would tell us that in order to be good defensive players we needed only to remember the three Ds: Delay, Direct, Destroy. Coincidentally, a similar axiom exists for investors hoping to maximize profits from a risk-averse portfolio: Diversity, Diversity, Diversity. Resource allocation, or where you choose to place your money, is one of the most fundamental cornerstones of investing, and rightfully so. Of course, certain types of investors or speculators have optimal portfolio compositions which are heavily skewed, and that works for their style. That being said, the average investor who is simply investing to grow their nest egg may be interested in the less risky investments.

At first blush it might seem like common sense. If your portfolio is composed solely of Nike, Under Armor, Lulu Lemon Athletica, and Adidas....
but of course, this is making a pretty big assumption that all of these companies essentially covary perfectly. We're assuming that when the whole athletic wear sector goes up, they all go up by the same number of points, and that when one has a successful period, they must naturally be taking an equal amount of value from their competitors. In addition to this concern, we should also consider more diversified companies, holding companies, and any other assets whose impact on a portfolio is not necessarily obvious. For example, how much do Amazon and Google co-vary? Facebook and Google? And now that Google is developing self-driving cars, are they in competition with the auto industry too?! When you take some of the projects at many companies into account, it can become overwhelming. Thus, in order to relax some assumptions and to account for complex contributions to a portfolio, we can use a number of statistical tools, the primary and most simple being a correlation analysis.

Correlation describes, on a scale of -1 to +1, the change in two stock's prices or one stock relative to an index (SP500, DJIA, etc.). In terms of readout, a +1 represents perfectly positively correlated stocks, a -1 would be a perfectly negatively correlation, and 0 indicates no correlation. In the case of constructing your portfolio, stocks with very little correlation to stocks you already own, as well as a stock with negative correlation to your other stocks are lend to the overall diversity of a portfolio. Perfectly positively correlated stocks do not complement each other and as a result provide zero diversity to a portfolio. Calculations such as these help investors to determine the diversity of a portfolio, but ultimately much more information and expertise go into determining the weighting of different sectors in a portfolio, as well as what price is a good price for a stock you are looking to add (asset pricing analysis is discussed in my previous post, if thats how you feel like spending the next 10 minutes of your life)

A Stock Solution; Asset Pricing Theory

Alright ya'll it's about to get stuffy in here. Something I have strong interest in is global markets and the stock exchange (well, strong for a biochemist with no economics training past high school). Naturally, this sector is perfect for exploring the widespread use of statistics and the unique and powerful ways they can be applied. While disciples of Benjamin Graham will warn that the valuation of a company's stock is not 100% tied to potential for an upward trend, this undoubtedly plays some role in the type of trading done on the market today. More than 50% of trading on the NYSE is done via something called High-Frequency Trading, which uses complex algorithms to buy and sell securities on the millisecond time scale for an overwhelming addition of small differences, resulting in a large profit for the companies employing these buying and trading algorithms. I'm not nearly savvy enough to describe these algorithms with sufficient detail, but I do want to discuss some aspects of these algorithms and other probability/statistics related applications in stock trading.

Investors deal with an immense amount of data, and more is generated every couple of milliseconds. As a result, a combination of instinct, experience, and sound statistical models are the professional trader's bread and butter. Various statistical tests are utilized to assess risk and confidence.

The first technique (and arguably the most central) application of statistics in stock trading is in Asset Pricing Theory. This branch of investment theory uses the calculated effects (effect size!) of various macro-economic factors, or the behavior of theoretical indices (indices track many different stocks, or sectors and are like a mean value representing how a sector is doing. Common examples are the S&P 500 or the Dow Jones Industrial Average) to forecast the expected return of an asset. I realize that all sounds a bit vague, so let's focus on a specific example:

We are all familiar with the T-statistic (departure of a parameter from its notional value and its standard error) which we use in Student's T-Tests. In the case of investing, the utility of this statistic is almost exactly the same as when we would use it to compare means. In fact, one of the foundations of investment statistics is formation and testing of a null hypothesis. In Asset Pricing Theory, the null hypothesis would propose that "the expected return of the asset is not different from the risk-free rate of return". In other words, they compare the asset in question to the performance of risk-free investments like some bonds or savings accounts. Given the historical returns of an asset and the risk-free investment of choice, an investor may find a T-statistic describing the difference between their asset (our sample of interest) and the risk-free investment (background, WT, negative control, etc.). Investors will then use the T-statistic as an indicator of the probability of observing the asset's returns under the assumption of the null hypothesis. Another similarity is that, in this branch of economics, they set their statistically significant p-value as 0.05, and utilize confidence intervals to have a better understanding of how this asset is likely to behave. Similar to our experiments, these calculations also rely on a certain "N", where a single N could be individual transactions involving this asset (in this case, higher volume stocks would be advantaged, due to a greater number of values), but this is not always the case.

This type of statistical testing is vital to many stock analysts, who will use the likelihood that a stock will perform better than a risk-free investment as part of their assessment of how to score the stock (what they should advise their advisees or their firm to do with regard to the stock). It can also indicate when a stock may be "overpriced" or "on sale". In many trading circles, the direction that a stock is likely to go short-term is less important than the absolute value of the company and what a stock of that company should be "worth", hence "Asset Pricing Theory".

Wednesday, April 27, 2016

The Lesser People

In last Thursday’s class, LeBreque brought to our attention many unethical research with groups of people considered to be "less important” by the researchers conducting the experiments. Although we have heard countless times about the Tuskegee Syphilis Study, Nuremberg Trials, Willowbrook Hepatitis Studies, Jenner’s vaccinations, among many others, we still must be reminded about these cases, which in turn remind us about the importance of seeing each life as equal when conducting research and the repercussions to minority groups if bioethics principles were not implemented. Even though we hear most about the cases stated above, it is clear these are not the only cases of unethical research and it is only normal that my mind trails off to things done to the people in my country. Listening to these cases made me remember some of the many unethical studies done in Puerto Rico by researchers who considered us to be an expendable population or in some cases, a population worthy of eradication.

One of the most shocking cases is that of Dr. Cornelius Rhoads, who in 1931 deliberately infected Puerto Rican citizens with cancer cells, killing 13 of the patients. Dr. Rhoads once said in a written document: “The Porto Ricans are the dirtiest, laziest, most degenerate and thievish race of men ever to inhabit this sphere… I have done my best to further the process of extermination by killing off eight and transplanting cancer into several more… All physicians take delight in the abuse and torture of the unfortunate subjects.”  He then went on to be in charge of chemical warfare projects and form part of the United States Atomic Energy Commission.

In the early 1950s, Puerto Rican women were used for experimentation in the making of the first birth control pill invented by Dr. Gregory Goodwin Pincus. Since laws in the U.S. did not permit full-scale experimentation, in 1955 Dr. Pincus and his colleague, Dr. John Rock, decided Puerto Rico was a perfect place to test out their pill due to the lack of anti-birth control laws. The trials quickly moved throughout the poor sectors in the island. The experiment was based on poor and working class women; these women were not told the pill was experimental and were not told the negative effects the pill could have on them. Three deaths occurred among patients who were taking the birth control pill. However, these deaths were not reported to be linked to the trials, despite strong circumstantial evidence that the pill was causing these unexpected deaths. It is believed it was also used as a form of population control to contain the poor sector.

Even though I have only gone into detail about two unethical experiments done in Puerto Rico, there have been many more, as would be the use of imprisoned Puerto Ricans as subjects to radiation experiments, sterilization policies, and the testing of the agent orange before its use in the Vietnam War. Even though some of the studies done to Puerto Ricans and other communities yielded important results, the use of subjects without their consent or their full understanding of the studies is unethical and no one has the right to grant more value to one life over another.

GWAS meets political forecasting

This is a really interesting approach to political forecasting. It is also a really great description of  how genome wide association studies work.

Google wide association studies

Tuesday, April 26, 2016

What is considered Ethical Research?

The lecture last week had me questioning what should be done with unethical research. My gut reaction is to dismiss it, after all, use of such research would on some level condone it. It could act as a guide post for morally questionable scientist to cross ethical lines in their own research in a sort of ends justify the means mindset.When we look at the information gathered from experiments such as those done in Tuskegee or Willow-brook it becomes easier to refute those studies. We can easily dismiss such studies because of the methods of acquisition was clearly morally bankrupt. However, taking one step back we must define what is or is not ethical.

Edward Jenner is hailed as developing the first vaccine when he used cowpox to protect against small pox infection. Many immunologist hold him in such high esteem for his accomplishments, indeed, his research is one of the main reasons we have vaccines today. However, when one looks closer at the methods of his seminal finding becomes questionable. He  inoculated a child with a known pathogen without knowing the effect it would have on the child. His first stroke of luck is that nothing severe occurred to the child. Next he purposefully injected the child with small pox in order to test his theory that the cow pox infection would be effective in generating protection to small pox. Luckily everything worked out and Jenner is praised for his contribution.

But was it a sound ethical experiment? He got extremely lucky that his experiment had no immediate or long lasting  detrimental effects on the child, or the other 23 subjects he tested his theory on. Had the experiment failed Jenner's name, far from being praised as it is now, would be used as a cautionary tale of a scientist caught up in his own hubris. If any scientist today were to attempt to replicate that experiment using the same methods, going off a theory based on limited observation, they would be looked at as the definition of inhumane. Using children as the primary subject base (not to speak of the ability of the child or their parent to give knowledgeable consent that isn't coerced), deliberately infecting said children with two different pathogens, with one known to cause severe illness. Not to mention the need for a negative control group which would receive pathogen without receiving the potential therapeutic. Finally the need to have enough subjects to achieve statistically sound study would require quite a few children. In Jenner's time, those methods were thought to be acceptable.

This just highlights that what is or is not ethical changes with the times. What was acceptable then is not acceptable now. If we accept this to be true then one must wonder what the future of ethical research holds. The values of society change and what is considered normal, just, and right changes with the times. Who knows, 100 years from now the society at that time could look at the research we're doing now and declare it unethical and inhumane. Maybe animal rights will become more prominent and we will move away from vertebrate animal studies. As for all the research up to that point will the society of the time completely disregard the data? Or will they use it but vow to never repeat its methodology? Will they wonder how our research methods could be so callous and barbaric, or will our research be looked upon with praise as Jenner's is now?

Did you know that research ethics can be dangerous?


The following blog post is in reference to the MMR research performed by Wakefield et al. 

 This is probably the most classic case of a piece of research going wrong. I use the term ‘research’ in the previous sentence very loosely. This is a classic case of fraud that went too far. Basically, the work was done by a now former researcher named Andrew Wakefield. He put together a bunch of data that showed a link between MMR vaccine and a new form of Autism. All of the data was proven to be false and Wakefield lost his license to practice medicine and is no longer a researcher. I would like to use this blog post to comment on some of the more interesting aspects from this case of deception.
One of the most interesting facts of this case is that it was known that the research results were not very good. The study used a very small cohort and it seemed impossible to generate the results that were published in the paper. Even the editor at the time thought that the work seemed very weak. This work was published in the Lancet, which is a pretty good journal. I think this is another good point, top journals make mistakes as well. Although, I have had a few good laughs reading some of the open access journals recently (I’m looking at you PLOS ONE). Of course the most damaging thing that came out of this entire ordeal is the effect it had on the general public. Most work that is published may be retracted for dishonesty but that is where the issue ends. The paper is retracted, researchers shamed and the world keeps spinning. But this case led to a new breed of stupid : the anti-vaccine movement. A group of individuals content with the idea that vaccines are bad for their children and thus vaccination must be avoided at all costs. Vaccine, although with much room for improvement, are probably one of the most important scientific achievements in history. No BS, I actually believe this. The number of lives saved by vaccination s staggering. The fact that most vaccines are made at a low price point has made them available to a majority of the world and led to the successful eradication of several diseases such as small pox. Think about that for a moment. A cheap and efficient method of eradicating a disease agent that would kill tens of thousands of people annually. But with the anti-vaccine crowds they don’t care for this history of events. So what do they do to justify their position? They reference the Wakefield study. This is why the Wakefield study was so damaging. People still believe it even though it is completely and terribly incorrect. So, in the end this study really did damage in the public opinion arena. This work obviously is cherry picked to prove an erroneous opinion and is the major reason why this study never should have been published. I never knew that a journal article could turn out to be so dangerous.

Why not? Nobody said we couldn't do it this way
The study originally started out as the Tuskegee Study of Untreated Syphilis in the Negro Male. The purpose of the study was to bring enough evidence forward to justify treating African Americans with syphilis infections. I think that is kind of strange. Why would you not treat people that have an infection that you know is bad for their health. Unless you are assuming that people of different races have different outcomes from disease because the races are so far apart from a genetic level? 

The study began in 1932 and originally involved 600 subjects. About 400 with disease and 200 without the disease. The men in this study never received informed consent and were given promises of free meals, free medical exams (yeah, I bet) and burial services would be paid for. These people were also lied to about the status of their disease. They were never told that they had syphilis. They were only told that they had ‘bad blood’. This is a phrase used to describe a number of ailments. In 1947, with Penicillin becoming the drug of choice for syphilis, the subjects were still not offered treatment. The study was to be conducted for 6 months but ended up lasting for 40 years.
I think that this case illustrates a number of ethical issues involving medical research. First, is the issue of do no harm. In clinical studies today it is common to give drug to an untreated or alternative treated group the drug under study if the data on the new drug is sufficient to prove that the treatment is significantly better at treating the disease. It is hard to understand why the participants would not be given the drug of choice in 1947. What is also bizarre is that the study in general makes absolutely no sense to me. It was already understood that syphilis was a bad disease so what was the point of the study? I think it is a great example of something that happens in science all too often, but certainly not to the degree of this study, which is basically putting a study together because you can. The lack of regulation in this study is staggering and makes me wonder what things would be like know if science was not heavily regulated. With the advancement of science over the last 50 years there are certainly experiments on certain researcher’s minds that would most likely put the Tuskegee study to shame.

The final outcome for the Tuskegee study was a settlement with the victims’ families. Also, study regulations were extremely modified and set the standards that we currently carry for medical research today. It should also be noted that the Tuskegee experiment did not become public knowledge until the media picked up on the issue. Could this study have continued for another 20 or 30 years without this coverage? Maybe. Overall the Tuskegee study is a primary example of why bioethics are so important for scientific investigation. It is important for keeping study participants safe but it is also good for science itself. What I mean is that it is important for science to be trusted by the general public. Otherwise you end up with stupid people not wanting to vaccinate their children.