Stories
Slash Boxes
Comments

SoylentNews is people

posted by martyb on Tuesday May 21 2019, @05:27AM   Printer-friendly
from the will-this-be-on-my-transcriptome dept.

In 1996, the an early genomic study found a correlation between variants of the SLC6A4 gene and clinical depression. For years after, researchers wrote paper after paper describing possible mechanisms and more detailed relationships of the gene and depression.

The Atlantic has an article about how a 2005 study using better methods and a larger sample set found no correlation, but researchers continued to treat the original (weak) correlation as a valid basis for further study for years later.

"You would have thought that would have dampened enthusiasm for that particular candidate gene, but not at all," he says. "Any evidence that the results might not be reliable was simply not what many people wanted to hear."

While this may not be the only case of suspect evidence leading to mountains of papers of dubious quality in scientific history, it's certainly a very modern one that raises the question of how far the replication crisis extends into the "hard sciences" rather than just the softer sort.


Original Submission

 
This discussion has been archived. No new comments can be posted.
Display Options Threshold/Breakthrough Mark All as Read Mark All as Unread
The Fine Print: The following comments are owned by whoever posted them. We are not responsible for them in any way.
  • (Score: 3, Disagree) by shrewdsheep on Tuesday May 21 2019, @03:58PM (2 children)

    by shrewdsheep (5215) on Tuesday May 21 2019, @03:58PM (#845845)

    Unfortunately, you still adhere to the old way of thinking. The optimism involved in longing for or forcibly deducing a biological mechanism has been thoroughly refuted by experience from GWASs. Together with publication bias, these are the two main mechanisms that have led to the utter failure of candidate gene studies in terms of reproducibility. BTW, there is always a plausible biological theory behind any association.

    Starting Score:    1  point
    Moderation   +2  
       Insightful=1, Interesting=1, Disagree=1, Total=3
    Extra 'Disagree' Modifier   0  

    Total Score:   3  
  • (Score: 5, Interesting) by ikanreed on Tuesday May 21 2019, @04:17PM (1 child)

    by ikanreed (3164) Subscriber Badge on Tuesday May 21 2019, @04:17PM (#845850) Journal

    I'm sorry, it sounds like you might possibly be sarcastic, but I'm still taking you seriously.

    GWAS studies I've reviewed in depth are just garbage. Finding with their main sample set the ability to explain 40% of variation in a phenotype, then a "secondary test" the same genes in a second population those shrunk to explaining about 3% and they still called it a positive study, with the 40% number in the abstract.

    Mind you, I work in a genomics company and would stand by the analyses we do here, but a lot of the original research done in the field is bad, and requires all the grains of salt in the Dead Sea.

    • (Score: 0) by Anonymous Coward on Tuesday May 21 2019, @06:01PM

      by Anonymous Coward on Tuesday May 21 2019, @06:01PM (#845882)

      Both types of study are utterly pointless. Every gene is correlated with every other gene and every behavior/etc, then they just filter out 99% of the correlations with an arbitrary significance threshold.