Researchers employ statistical techniques to condense and simplify the data while maintaining the majority of the important information in order to make the data more manageable. PCA (principal component analysis) is perhaps the most widely used approach. Imagine PCA as an oven with flour, sugar, and eggs serving as the input data. The oven may always perform the same thing, but the ultimate result, a cake, is highly dependent on the ratios of the ingredients and how they are mixed.
Written by
cesces
To view profiles and participate in discussions please or .
"Genomic testing looks at the ways in which your genes interact and what those interactions mean to your health. Genomic testing is different from genetic testing because it looks at all of your genes, rather than detecting a single gene mutation."
It would appear to me that most of what we discuss here relaying to prostate cancer is about genetic testing (for single genes) as opposed to genomics?
Genomics looks at which genes are or aren't expressed. Genomics can be germline (Color Genome Dx, etc), somatic (Decipher, Foundation one, Caritas, etc), or both (eg, Guardant 360). The genetics studies they are talking about have to do with the occurrence of genes in an ethnic or geographic population. It has nothing to do with individual diagnosis.
Hi All, Data Scientist here. PCA is a technique to look for “linear relationships” among variables ina data set. For example, if you had a set of data describing various recipes for rice we could use PCA to see if there is a fixed relationship between the amount of water and rice to be used and how well that ratio describes all recipes. If we only have one kind of rice, consistent high quality measures and only one style (sticky, parboiled, etc…) we are trying to cook the PCA would describe that discovered ratio very well. If our data mixes various types of rice, target styles and has a mixture of measurement accuracies then PCA will still describe average ratios correctly but will have must less accuracy and therefore value. Beyond this, the well known problem with PCA is that If relationships cannot be well described by fixed ratios, for example rice cooked in larger vessel has differing water/rice ratios (it does), then PCA will still produce results but they will not be useful or predictive.
Nevertheless many academic studies in many many fields begin with PCA analysis because if relationships are largely fixed ratios PCA runs fast and works on incredibly big data easily. Without digging further I say the article is a rather “click bate-y” in that it is pointing out well known problem and just using a search engine with the combined terms “Genomics” and “PCA”, finding thousands of results and then claiming that huge swaths of research are invalid is unfair to the thousands of projects that just use PCA as an obvious starting point nd move on to more complex methods as merited.
the article refers to use of PCA for population genetics. Relevant to PC, we might examine patient ethnicity and use PCA to make a very simplistic model proposing that ethnic background had a fixed influence over prevalence of PC is related populations. We could use PCA to make a model given percent of Central African (positive contributes) and East Asian ancestry (negative contributor) to predict rate of PC. This model would be terrible because it would only use two factors and would assume that a 1% increase in African ancestry equals a fixed increase in PC risk. Actual risk increase is more likely a complex response curve and poorly modeled by a straight line. There may be some garbage studies that make such an obvious mistake. But I doubt it. Most studies would start with the simple model, add additional factors and use fancier techniques to derive a more precise risk curve related to population factors. The main claim of the article involves searching article databases for combine terms which in my opinion doesn’t really invalidate all the ‘hits’ most of which used PCA for initial exploratory analysis.
It’s the equivalent of searching food diaries and claiming everyone eats nothing but breakfast because the terms “toast and coffee” are very common. Lots of pleople who eat toast also have lunch and dinner
Cancer is not a genetic disease as the the Big Pharma & its subsidiaries FDA & AMA has been promoting, it is a metabolic disease, dysfunctioning of mitochondria.
Look for the books and videos of honest scientists, who hasn't sold their souls to Big Pharma, such as Prof Thomas Seyfried, Bruce Lipton Ph D, Dr Joe Dispenza, Dr Andreas Moritz, Dr Eric Berg, etc.
Trust your body's intelligence which is capable to heal anything. When you disregard your body, treat it like dirt, ignore its messages, surrender to profit oriented doctors and their beyond-doubt-proven, toxic, lethal so called solutions, then you have no way to heal yourself. Even a short lived relief is not going to prevent recurrence, or secondary cancer developing another part of your body, because of chemo, radiation and hormon therapies you consented.
Content on HealthUnlocked does not replace the relationship between you and doctors or other healthcare professionals nor the advice you receive from them.
Never delay seeking advice or dialling emergency services because of something that you have read on HealthUnlocked.
Sleep disturbances are underappreciated in prostate cancer survivorship
JAMA October 2108 - Does Complementary Medicine Increase Risk of Death for Cancer Patients?
The human microbiome and its link in prostate cancer risk and pathogenesis
Coffee and Tea and PC
Castrate Resistance Mechanism & Statins
