[Goose]

I think we're mostly repeating ourselves here. I won't convince you, that's fine.

There is plenty of evidence the above statement is false, as I have pointed out previously. How many people read the CELL paper? Presumably a very large number. How many of them detected the problem? Zero. As far as I know nobody questioned the figures until the computer program pointed out the anomalies. That is pretty good empirical evidence that the software was "necessary" to detect the anomalies. It may be that once one knows there is an anomaly, they may be able to detect it. However, at that point they are looking for a known anomaly, which is much easier.

A random reader may not be able to easily determine a figure is altered, but that's a different question than whether the senior author can determine that. The senior author should be very aware of the underlying data or involved in the experimental collection of that data.

That doesn't mean they need to have seen every blot or immunostain the lab has done. But if it's going in your Cell paper, you should have seen these experiments replicated in your own lab and seen the photos of the gels or blots that will make up those figures.

You are assuming that the underlying data and the photos contradict what the published photo shows, a fact not in evidence. We don't know that is the case. The question of how many times this experiment was replicated is a valid one. I don't know the answer. If it was an N of 1, that is another issue, and that is a valid question. If it was a confirmatory experiment to supplement other data, it is less of an issue.
It is quite possible that MTL did see the photos but that they had already been manipulated. It is also possible that the photos showed the same thing before manipulation. All he ever saw was what he expected to see. Assuming the postdoc did the manipulation, I would not expect that they presented data to the PI that clearly refuted what later became the conclusion of the paper and then later altered the images to change the result. That would be crazy, and MTL would indeed know that the data had flipped 180 degrees. Unlikely.

Close familiarity with that data would have enabled an observant PI to catch the manipulation at issue.

Again, you are assuming the manipulated photo was "significantly" different than the ones the PI saw that were allegedly not manipulated. They may not have been materially different. Given the pattern in the other papers it could be there was a cosmetic manipulation that was done for publication that did NOT contradict what MTL had previously seen. OTOH, if the postdoc was manipulating the data, she would have been crazy to ever show MTL a contradictory photo. That would be asking for trouble. If they were both engaged in a conspiracy to publish a bogus paper, it is a moot point as they both knew what was going on.

If the conclusion drawn from the data is in fact correct, there is every reason to believe that the manipulation did not change the photo in any significant way. If that conclusion is not correct there is every reason to believe the manipulation did in fact change the data.

If we have independent verification that might mean nobody did anything unethical.

Image manipulation of this kind even if the conclusion is valid is still very unethical.

Absolutely don't agree. Image manipulation for publication happens all the time. It is obviously unethical to do something that changes the results and character of the data. Contrast enhancement for instance is often applied. Edge sharpening is also often applied. This is often done in "modern times" right as the photo is made. Is "auto-focus" allowed? It is obviously a matter of degree and intent. I would agree that a statement detailing what techniques were used to "enhance" the image would be appropriate, but I can't get excited about something that is actually "cosmetic" not being cited.

[Goose]

I think we're mostly repeating ourselves here. I won't convince you, that's fine.

There is plenty of evidence the above statement is false, as I have pointed out previously. How many people read the CELL paper? Presumably a very large number. How many of them detected the problem? Zero. As far as I know nobody questioned the figures until the computer program pointed out the anomalies. That is pretty good empirical evidence that the software was "necessary" to detect the anomalies. It may be that once one knows there is an anomaly, they may be able to detect it. However, at that point they are looking for a known anomaly, which is much easier.

A random reader may not be able to easily determine a figure is altered, but that's a different question than whether the senior author can determine that. The senior author should be very aware of the underlying data or involved in the experimental collection of that data.

That doesn't mean they need to have seen every blot or immunostain the lab has done. But if it's going in your Cell paper, you should have seen these experiments replicated in your own lab and seen the photos of the gels or blots that will make up those figures.

You are assuming that the underlying data and the photos contradict what the published photo shows, a fact not in evidence. We don't know that is the case. The question of how many times this experiment was replicated is a valid one. I don't know the answer. If it was an N of 1, that is another issue, and that is a valid question. If it was a confirmatory experiment to supplement other data, it is less of an issue.
It is quite possible that MTL did see the photos but that they had already been manipulated. It is also possible that the photos showed the same thing before manipulation. All he ever saw was what he expected to see. Assuming the postdoc did the manipulation, I would not expect that they presented data to the PI that clearly refuted what later became the conclusion of the paper and then later altered the images to change the result. That would be crazy, and MTL would indeed know that the data had flipped 180 degrees. Unlikely.

Close familiarity with that data would have enabled an observant PI to catch the manipulation at issue.

Again, you are assuming the manipulated photo was "significantly" different than the ones the PI saw that were allegedly not manipulated. They may not have been materially different. Given the pattern in the other papers it could be there was a cosmetic manipulation that was done for publication that did NOT contradict what MTL had previously seen. OTOH, if the postdoc was manipulating the data, she would have been crazy to ever show MTL a contradictory photo. That would be asking for trouble. If they were both engaged in a conspiracy to publish a bogus paper, it is a moot point as they both knew what was going on.

If the conclusion drawn from the data is in fact correct, there is every reason to believe that the manipulation did not change the photo in any significant way. If that conclusion is not correct there is every reason to believe the manipulation did in fact change the data.

If we have independent verification that might mean nobody did anything unethical.

Image manipulation of this kind even if the conclusion is valid is still very unethical.

Absolutely don't agree. Image manipulation for publication happens all the time. It is obviously unethical to do something that changes the results and character of the data. Contrast enhancement for instance is often applied. Edge sharpening is also often applied. This is often done in "modern times" right as the photo is made. Is "auto-focus" allowed? It is obviously a matter of degree and intent. I would agree that a statement detailing what techniques were used to "enhance" the image would be appropriate, but I can't get excited about something that is actually "cosmetic" not being cited.