5 ms·

The entire opening of the article discussing 2020 papers erroneously using skin cancer data in breast cancer research papers which was a known error for at leas

The entire opening of the article discussing 2020 papers erroneously using skin cancer data in breast cancer research papers which was a known error for at least 13 years - makes the point very strongly that " ... but these studies/experiments have shown that understanding to be incorrect. Now we believe this." happens much slower than perhaps people expect. I understand science "is expected to be wrong" sometimes, but I'm still somewhat boggled that there's 13 years worth of (presumably) peer reviewed papers making claims about breast cancer using "wrong science" that was proven wrong in 2007... (I wonder what _else_ all those peer reviewers let slide or didn't know when reviewing other papers?)

dayiel 6y ago
Actually, the cell line issue might not be so bad or rather all cell line work is much worse than you all imagine. No cell lines actually resemble the real tissue in gene expression if you say, do an rna screen and see what actuallg gets expressed. They are changed by line establishing process, use of various media, number of passages aka how long they’ve been in culture from thawing a line till the test (cant be too little or too much, requires line to recover after being frozen), person who’se culturing (some people tend to starve cells and passage full plates and have different parameters of work even despite some labwide protocols), adherence to the plate and growing in flat monoculture, getting plates out too much or watching them too lonv under the microscope. Some effects are temporary and not so serious. Some effects are rather serious. Like the fact that flat surface cell lines will never be a good proxy for the actual tumor growing in 3d in the actual human. That’s why people experimenting with spheres, multucultures with other cells, organelle cultures, animal models, and eventually all the clues from all the models add up to something that could be a theory of some mechanism. And then in most cases when applied to humans in real life it might not work or be harmful. Now, the skin cell in cell media was likely arrayed and had features similar to breast cancer cells so it might have been okay to use for this super imperfect approximation. Not sure if this particular instance. In vitro is only a part of it and peer reviewed journals differ in their authority ratings, with big boss journals like cell requiring you do deliver crazy amounts of different models to make any statement. Reading those papers is like reading house ads, scientists kind of learn to cut through bs
- d110af5ccf 6y ago
  I agree with the sentiment expressed here. > Reading those papers is like reading house ads, scientists kind of learn to cut through bs In addition to recognizing general BS, in any given field the papers come with an enormous block of implicit fine print. An important part of being an expert in a particular field is knowing what that fine print is.
aspaceman 6y ago
13 years for turnaround on bad results isn’t abysmal in my view. Science is hard and will take a long time. Getting people used to that seems impossible though.
- d110af5ccf 6y ago
  And this, of course, is why double blind clinical trials are so incredibly important.
rbanffy 6y ago
> I wonder what _else_ all those peer reviewers let slide or didn't know when reviewing other papers? Peer-review is only one layer for identifying errors. Some things will pass through it. Then there will be retractions if the paper proves to be flawed, but, for that to happen, someone needs to identify the flaw and reach out to the authors. It's a slow process.

dayiel 6y ago

Actually, the cell line issue might not be so bad or rather all cell line work is much worse than you all imagine. No cell lines actually resemble the real tissue in gene expression if you say, do an rna screen and see what actuallg gets expressed. They are changed by line establishing process, use of various media, number of passages aka how long they’ve been in culture from thawing a line till the test (cant be too little or too much, requires line to recover after being frozen), person who’se culturing (some people tend to starve cells and passage full plates and have different parameters of work even despite some labwide protocols), adherence to the plate and growing in flat monoculture, getting plates out too much or watching them too lonv under the microscope. Some effects are temporary and not so serious. Some effects are rather serious. Like the fact that flat surface cell lines will never be a good proxy for the actual tumor growing in 3d in the actual human. That’s why people experimenting with spheres, multucultures with other cells, organelle cultures, animal models, and eventually all the clues from all the models add up to something that could be a theory of some mechanism. And then in most cases when applied to humans in real life it might not work or be harmful. Now, the skin cell in cell media was likely arrayed and had features similar to breast cancer cells so it might have been okay to use for this super imperfect approximation. Not sure if this particular instance. In vitro is only a part of it and peer reviewed journals differ in their authority ratings, with big boss journals like cell requiring you do deliver crazy amounts of different models to make any statement. Reading those papers is like reading house ads, scientists kind of learn to cut through bs

d110af5ccf 6y ago
I agree with the sentiment expressed here. > Reading those papers is like reading house ads, scientists kind of learn to cut through bs In addition to recognizing general BS, in any given field the papers come with an enormous block of implicit fine print. An important part of being an expert in a particular field is knowing what that fine print is.

aspaceman 6y ago

13 years for turnaround on bad results isn’t abysmal in my view. Science is hard and will take a long time. Getting people used to that seems impossible though.

d110af5ccf 6y ago
And this, of course, is why double blind clinical trials are so incredibly important.

rbanffy 6y ago

> I wonder what _else_ all those peer reviewers let slide or didn't know when reviewing other papers? Peer-review is only one layer for identifying errors. Some things will pass through it. Then there will be retractions if the paper proves to be flawed, but, for that to happen, someone needs to identify the flaw and reach out to the authors. It's a slow process.