poorgradstudent, I like you and value your comments, but you are way off base here. You do a dis-service to the board, and to academic scientific researchers with your comments.
{{I don't think it is a major over-estimate.
I'm saying you could look through 100 papers and find simple flaws in at least half of them that make the conclusions untenable.}}
I think you are greatly over-estimating the kinds of errors that are made and the impact they have. It is also a function of the project and the organism specific as different tools are utilized and different analyses have very different levels of complexity.
It appears to me in this thread that some people are placing researchers in industry on a pedestal of virtue, and acting as if they don't make errors, or aren't driven by pressures to find results that are significant. That is completely false. I worked in both and know it for a fact. You might want to say there are errors in both technique and in understanding techniques academic research and pharmaceutical in house research that share many of the same flaws you list below.
{{For example: 1) Check the sheer number of Western blots that show small windows for the signal, and you can see that they've cut out the "nonspecific" signals. Well, if you speak with many of these people, the only reason the signal they chose is "specific" is because it is near the right size. That's demonstrable BS and goes on all over the place. Given the prevalence of this technique and people's refusal to validate the antibodies they use, that's a giant source of error that would invalidate many studies.}}
The trimming of blots to show only the band size range of interest is far too common, and yes, fraught with issues. I think with the advent of supplemental data, one could place the whole blot there for perusal. But even that is problematic as where do you put the line. We always run smaller proteins off the gel to get resolution for the size range we expect our proteins to run, or may not have resolving power for very large proteins.
Now I work mostly in the budding yeast Saccharomyes cerevisiae. I and others epitope tag proteins so what you are talking about isn't relevant. We use different sized tags to shift the mobility based the different tag sizes to make it unambiguous that we are looking at the correct protein. Despite the power and ease of epitope tagging, I made antibodies to different yeast proteins and used differentially tagged proteins as my reference standard. I also pre-screened my rabbits to select those which lack antibodies which cross reacted to bands anywhere near where my proteins run. I made these reagents because of issues have been cropping up with some tags in different mutant strain backgrounds, as well as in cis, when as generate new mutants of my proteins of interest and tried to tag them. I have sent these reagents out to many labs, and provide advice and caveats they need to be concerned with including Ab concentrations for blots and immuno-fluorescence.
I agree that antibody validation is crucial, but again you resort to a bit of hyperbole. Even in mammalian cell studies, the advent of siRNA enables one to show that the band in the small zone you mentioned above is diminished after siRNA knock-down. You see scientists use this technique to validate their Abs in many or even most papers, especially for the critical proteins central to their hypothesis. Mammalian researchers also employ tagged proteins transfected into cells differentiate their transfected mutant proteins from endogenous proteins as the Abs recognize both forms, the putative endogenous WT and the retarded mobility tagged protein.
I am not sure where you were trained but the errors and flaws would never be accepted in my lab, in the labs where I have joint meetings, or institutions I have been as these are often exposed in internal seminars and meetings. I am also harsh in the papers I review to weed out and correct potential flaws prior to acceptance for publication. Do errors still get made and inaccurate interpretations get published? Sure but to call 50% of all published work wrong or highly flawed is absurd.
I actually resent your broad characterization of fraud and or gross incompetence as it is very misleading and unfair. You also provide fodder for the imbecile conspiracy theorists and anti-science luddites who want to cut science funding further, which would only serve to exacerbate problems.
News 
Market Data 
Discover 
AI
