Vụ giả mạo dữ liệu khoa học mới

Nguyễn Xuân Hưng

Administrator
#1
Vụ này liên quan đến bà giám đốc ở chỗ cũ của mình! Nếu không có thời gian đọc hết thì xem các hình ảnh phía dưới thôi.

Vụ này đáng đi vào lịch sử vì tính hệ thống, những hệ lụy xung quanh.


http://www.martinfrost.ws/htmlfiles/aug2010/rothwell-precis.html

Trích 1 đoạn:

The Weibauer effect
The chickens came home to roost when Dr. Weibauer began contacting the journals. Her criticisms included not only re-use and the cunning manipulation of data but the impossibility of obtaining PCR fragments that could never be amplified given the published primers, and the unnatural consistency of the error bars. Here are but a few examples that were perpetrated by the Budagian/Bulanova/Paus/Bulfone-Paus fraudsters that can be found in the scientific literature:
Bulanova et al., J. Immunology, 167:6292-302, 2001

This is taken from figure 8 of Bulanova et al., 2001.

First look at the top poanel. You will see that in lanes 2 and 6 there is a higher molecular weight band. In all lanes there is a much darker lower molecular weight band.

Now go to panel 4: this panel is the same as the top panel except that the lower molecular weight bands have been cut off
I understand that a correction is to be placed in the journal. But can it be believed? It will besaid that the cropped section of the top panel of figure 8 appearing in the 4th panel of figure 8 was a “slip” of the paste key.

A simple inspection of figure 8 shows that the “cropped” portion given in panel 4 and the top panel completely overlap. How did this “slip” slip Professor Doctor Doctor Bulfone-Paus’s mind until it was pointed out by Dr. Weibauer?

This “slip” was made to make an important point in the paper. Other things which obscure what is going on in this paper are figures 11 and 12, which are of such poor quality that they cannot be interpreted.

This is taken from Figure
4A of Bulanova et al.,
2003.

Look at the two highlighted bands in lanes 2 and 3 of the middle
panel.

The higher molecular-weight band
looks decidedly like the lower molecular weight band.

Now you have to compare these two bands with the same two that
are highlighted in Figure 9A of the same paper.








In Figure 9A the same two
bands found in Figure 4A
are in 9A.

The high molecular weight band is also surrounded by
aberrations that are the result of “cut-and-paste”.


How did the journal miss
this?

Budagian et al., EMBO J., 24:4260-70, 2005

The EMBO paper is, in many ways, the pinnacle of Budagian-Bulanova-Paus-Bulfone-Paus’
achievements leading to the press hailing professor Bulfone-Paus as a star of German science.

The title of the piece contains the phrase “promiscuous liaisons” which, we are told, was coined by Dr.
Budagian in honor of Professor Doctor Doctor R. Paus.


This is taken from Figure 4B of Bulanova et al., 2003. Have a look at lanes 1, 2 and 3 which is under the label IL-15Rαδ4.

You’ll see in lanes 1 and 3 there is a noticeable band at about 37 KDa, in lane 2 there is a noticeable band at about 32 KDa.

If you take this section as a whole (all encompassed by the red box) you will find it in lanes 1, 2, 3 under the heading IL-15αδ3,4,5. It has been transposed "en bloc” to the region of about 20 KDa.

Also, notice how the area between about 20KDa and 37 KDa 1, 2, 3 under the heading IL-15αδ3,4,5 is strangely smooth (likely the join), except for the large, very dark round blob in lane 2 (almost looks like it’s added later).
This is taken from Fig. 2A Budagian et al., 2005.

Look at the third row. Different IPs are supposed to have been done to the samples in the left and right panels, yet the right panel looks like a five-fold exposure of the left panel.

Please note the imperfections which are the same in the lanes of both panels:-


1. The dark left extremity in both control lanes.

2. The upward inclination of both right lanes going from left to right.

3. The upwards tail in the right extremities of both right lanes of the two panels.

4. The upward tail is also seen in both middle lanes of both panels, but is less pronounced.

5. The lower fainter bands near the bottom of both panels also look like copies of each other in the different panels.



The simplest hypothesis which fits the data is that the right panel is an overexposure of the left panel.
This casts doubt on the rest of the data

This is taken from Fig. 1C Budagian et al., 2005.

The consistency of the error bars should be matched in the consistency of the primary data.
We need to see the primary data and a statistician needs to analyze the original data.


This is taken from Fig. 2B Budagian et al., 2005.








As with Fig. 1C, the consistency of the error bars should be matched in the consistency of the primary data.
We need to see the primary data and a statistician needs to analyze the original data.



This is taken from Fig. 7E Budagian et al., 2005.


As with Fig. 1C and 2B, the consistency of the error bars should be matched in the consistency of the primary data.

We need to see the primary data and a statistician needs to analyze the original data.




This is taken from Fig. 7E Budagian et al., 2005.







As with Fig. 1C and 2B, the consistency
of the error bars should be matched in
the consistency of the primary data.

We need to see the primary data and a
statistician needs to analyze the original
data.



Figure 3B DATA RE-USE

This is taken from Fig. 3B of Budagian et al.,
2005.

Look at lane 6 in the first panel (encompassed by a red box). You will see two bands at the bottom and a faint one at the top.

Now look at the 4th (bottom) panel, lane 6 (also encompassed by a red box).
The lower two bands looks like an over-exposure of the lower two bands in lane 6 in the top panel; the light band at the top of lane 6 of the top panel is also darker in lane 6 of the bottom panel.

All the imperfections are the same between the bands of the top and bottom panel. For example the right ends of both lowest bands is thicker, there is an upward “tick” to the right end of both these lowest bands.

The first panel is an IP followed by a WB. The last panel is a WB done on a lysate (no IP). The WBs should not be identical, for example in the imperfections.
(top)
Figure 1c error bars same we need to see original data This is taken from Fig. 1C of Bulanova et al., 2007.


As with Fig. 1C, 2B and 7E the consistency of the error bars should be matched in the consistency of the primary data.

We need to see the primary data and a statistician needs to analyze the original data.




Figure 3 -error bars and re-use of data!
we need to see the original data


This is taken from Fig. 3 of Bulanova et al., 2007.

This is a complex figure so I will try to go through it slowly.

In Part A what struck me about the dark bands in the lower panel (CM) of at 15 and 10 KDa is that they are so juicy and perfect as if they have been stuck on – it would be good to see the originals. One is, of course, reminded of the clumsy manipulation of Figure 9A in Bulanova et al., 2003. Now look at Part B, panel L.

The dark bands in lanes 2 and 3 are of similar shape and should be compared to those in Figure 4A (WB). In Part C, COS-7, of the values at 6 to 8, 11 out
of 11 error bars are essentially the same (there is a single much lower value which has a very small error bar).

In Part C, HeLa, 11 out of 11 error bars are essentially the same.

As we have said before the consistency of the error
bars should be matched in the consistency of the primary data.

We need to see the primary data and a statistician needs to analyze the original data.


In part E, the top and bottom panels of panel E seem to be different exposures, the top panel being an over-exposure, until saturation, of the bottom panel, so making them look like different shapes when they are not. The higher lighter smudges in lanes 2 and 4 may have been added.

They are of a different quality than the dark bands. We need to see the original data
(top)

Figure 4 - error bars and re-use of data

This is taken from Fig. 4 of Bulanova et al., 2007.

The error bars encompassed by the red boxes in panels A, B, and C appear the same. As we have said before the consistency of the error bars should be matched in the consistency of the primary data. We need to see the primary data and a statistician needs to analyze the original data.

In Figure 4A (WB) the dark bands in lanes 4 and 5 are the same shape and are the same shape as the dark bands in Figure 3A panel L, lanes 2 and 3.

(top)

Figure 5 - Error bars similar or identical
This is taken from Fig. 5 of Bulanova et al., 2007.



In Figure 5A, 4 out of 4 error bars we can see are the same.

In Figure 5B the 3 error bars we can see are nearly the same.

As we have said before the consistency of the error bars should be matched in the consistency of the primary data.

We need to see the primary data and a statistician needs to analyze the original data.



(top)

Figure 7 - Data re-use?
Duitman et al., Mol Cell Biol. 28(15):4851-61, 2008

This paper shows that the falsifying of data is endemic in the Bulfone-Paus laboratory.

They may not now know the difference between reality and falsified data. One correction has already been made, and published, to this paper in response to a Weibauer criticism but there are others that need to be addressed.

(top)

Data re-use? continued
The next image (Figure 5 of Duitmann et al., 2008) is disturbing because controls for this experiment were likely never done. As shown the CD81 lanes are spliced together. The originals need to be carefully examined and the contents of the original lanes determined with certainty. I have been told that Bulfone-Paus is confident that the investigating commission will not examine these data because it is lazy and focusing only on the Weibauer criticisms.



This is taken from Fig.5 Duitman et al., 2008

Are the bands seen on the controls from unrelated blot - i.e., they were never done. Did Bulfone-Paus force her co-workers to supress the evidence? The commission should investigate this thoroughly.

Will there be a “white-wash” by the committee or do they mean business? Time will tell and the World will be watching.
 
Bắt đầu có disillusion về nghiên cứu khoa học. Ở chỗ mình người ta ép oral gavage chuột 6 lần một ngày để hy vọng có đáp ứng miễn dịch với vaccine candidate đang thử nghiệm. Thằng cu bạn cho 40 con chuột ăn, mỗi con 6 lần một ngày. Đêm về gặp cả ác mộng :xinkieu:
 
Có một thằng cha gì ở Mỹ chuyên gia làm về chloroplast transformation. Ra hơn chục bài báo về cái này. Toàn lên báo đỉnh của plant như Plant physiology, planta, PNAS, Nature Biotech...v.v. Vậy mà không có một ai có thể lặp lại thí nghiệm của lão này. Có lần trong 1 bài báo recombinant protein lên tới 40% total protein:???:, hơn gấp mấy lần rubisco, protein phổ biến nhất ở thực vật.
 

Facebook

Top