On defecting

Jenny has a new shiny. It’s a device for imaging chemiluminescence–a standard procedure in any lab that works with proteins. The traditional way of doing this is on film, but it seems a lot quicker, safer and environmentally-friendlier to do it with one of the imaging gizmos.


Except I’m a little bit worried. I was reading a paper just now, trying to figure out how to summarize it for our Faculty Dailies, and came across this figure:


Now I have no idea how this image was obtained (the Methods section mentions neither film nor fancy-schmancy new devices), but either way that is one butt-ugly blot (BUB for short). I am worried that it is obtained with a FSND, because you really have to be a bit of an imbecile to get that level of pixellation when digitizing a blot by scanning a film. I wouldn’t ever want to publish something that looked like that–accusations of over-processing aside, it simply looks wrong.

Are we likely to see more BUBs as FSNDs gain in popularity? Is a whole way of life and aesthetic pleasure at stake here? Say it ain’t so, Jenny.

Say it ain’t so.

About rpg

Scientist, poet, gadfly
This entry was posted in Rants, Science. Bookmark the permalink.

20 Responses to On defecting

  1. Samantha Alsbury says:

    you really have to be a bit of an imbecile to get that level of pixellation when digitizing a blot by scanning a film
    It’s not impossible though – don’t blame the technology it’s the users!

  2. Richard P. Grant says:

    Hey Sam,
    I guess my question is, will the new tech make it easier for users to let their imbecility shine through? After all, look what Windows and the internet did for morons…

  3. Samantha Alsbury says:

    hehe, yes quite possibly…but at least we have peer review to pick up on these things!

  4. Stephen Curry says:

    I would say that the sin is not in the resolution (perfectly adequate for quantitation in this case), but in the omission of experimental details from the methods.

  5. Richard P. Grant says:

    The peer reviewers apparently were not concerned about either point. They’re not my peers, then.

  6. Ian York says:

    If you go back through the history of phosphorimagers, you’ll see a lot of early images that look like this; modern imagers turn out much higher-resolution images. The first phosphorimager I used, in the mid- to late 1990s, looked much like this. I suspect that this is an old machine, perhaps through an old paper.
    You need to separate aesthetic and scientific requirements here. As Stephen says, the gel may be ugly, but it’s probably acceptable for quantifying the images in this case. Should peer reviewers be concerned about pretty gels? Is beauty or truth more important?Admittedly I have had a couple reviewers comment about attractiveness — but sometimes you just can’t get beauty.

  7. Richard P. Grant says:

    Hi Ian
    nice to see you again. This is actually a very new paper.
    My major problem stems from the impression the figure gives: the pixellation is what you’d expect if something had been over-compressed, or over-manipulated. I’m not accusing the authors of fraud, but this is the kind of image that one might find suspicious.
    My secondary problem is that if you can’t/won’t get something that looks nice (I mean, it’s not an immunoprecipitation) then what does that say about your attention to detail with respect to the science?
    Alarm bells, they tinkle.

  8. Cath Ennis says:

    I did a ton of CAT assays during my PhD (so many that they came to be known as CATH assays), and I was sooooooo happy when we finally got a phosphorimager. It made things so much easier. (Sighs happily and nostalgically).
    My PhD supervisor was on the editorial board of a society journal, and emerged from his office one day to get the lab’s opinion on a BUB in a paper he was handling. We all said the same thing – “the second band from the left looks like it was drawn on in marker pen”. Neither of the reviewers had commented, but as editor, my supervisor requested the original film to check against the figure. The authors did not return his email, and the paper was never seen again…
    As the boss man said, “if you’re going to fake data, at least fake it well”.

  9. Richard Wintle says:

    That is some pretty nasty pixelation, yup yeah.
    I wonder that more people don’t simply take bog-standard digicam photos of output displays, or printout photos, or even X-ray films, rather than using fancy equipment to scan them. The average pocket point-n-shoot has excellent resolution as compared with a lot of lab devices.

  10. David Martin says:

    The problem isn’t with the pixellation (though it is ugly), it is with the quantitation, and the dynamic range of such scanners. I’ve seen them used since the early 90s (mostly for radiolabel work) and the same things still apply. What are the thresholds, what is the white point, black point, gamma setting on the curve? What bitdepth? was there a standard curve to quantify density against load? There is a lack of rigour in getting decent numbers from blots (personally I’d rather see the processed results with error bars than yet another picture of a band. That does presume people can run decent gels which seems to be just as lacking a skill now as it was when I was a lab rat rather than a data juggler).

  11. Richard P. Grant says:

    Very good points, David. If this was simply a ‘is it there or not’ it’d be different.
    I also notice a hint of something else in the lower left hand corner of that blot. I’d like to know what’s going on.

  12. Jennifer Rohn says:

    My guess would be molecular weight marker. When I was a lass we had to show the entire gel, including the markers. I think the fad for cropping arose circa 1996-7.

  13. Richard P. Grant says:

    Isn’t that fad being reversed now? I suspect it arose when people realized they could run Photoshop on their desktop Macs, and finally the journals are catching up.
    I it was JCB had a rant about this a couple years back, asking for the whole gel, no matter how messy, etc.

  14. Richard Wintle says:

    Heh. Props to David Martin for complaining about people not being able to run decent gels.
    When I was a lad (to steal Jenny’s intro slightly), I worked in a lab doing deletion analysis using Southern blots. I was taught that if the gel was done right, you didn’t need densitometry. This was looking for 0 vs. 1 vs. 2 copies though, so easier than quantitating some continuous variable.
    Regarding Photoshop, the first time I encountered it (on a shiny new Mac Quadra as I recall), my fellow grad student, just to see if could be done, scanned in a sequencing gel autorad and cloned a band to “make” a mutation. It was, as you might imagine, thoroughly convincing.
    And no, he never used such (stealing a word again) skullduggery in his publication career… as far as I know.

  15. Richard P. Grant says:

    Quantification does depend on the experiment. It was Rutherford who was so down on stats, wasn’t it?
    I can’t remember the first time I used Photoshop…although I do remember feeling a twinge of sadness that the days of Letraset were numbered.

  16. Richard Wintle says:

    I’d forgotten about Letraset… I used it once upon a time but quickly graduated to the “mock up the labels with a blank box in MacDraw, then glue the photograph into the box” approach. Some of the results of which are sitting not 8 feet away from me, gathering dust.
    My band-cloning colleague actually did his entire PhD thesis (circa 1995) with scanned figures embedded in the Word document, which seemed like NASA science at that time. He had to make each chapter a separate document and link them together – otherwise the computer would have choked on all the images in one monolithic document.
    Pretty sure the first time I used Photoshop per se was only a few years ago… for photo editing no less.

  17. Richard P. Grant says:

    Oh yes, I used that MacDraw method too, but getting the printer to print stuff in the right place always seemed troublesome.

  18. Kyrsten Jensen says:

    I’ve had BUB even with film (woops – did I move it? yup. Shouldn’t have had coffee when I’m trying to detect picograms of material).
    I have to say, the day the lab got a licor odyssey(http://www.licor.com/bio/products/imaging_systems/odyssey/odyssey_imager.jsp), I rejoiced! I was the tech in the lab who got to test it and enhanced chemiluminescence (we always did colourimetric until I came to the lab with my ECL mad skillz I picked up from a co-op term at BigPharma). For the first time to see how quantitative it could be. The cool thing is that with the LICOR I didn’t even have to use ECL to get good sensitivity – threw on a primary antibody that was PE-labelled, scanned the blot, and bands! I also used some GFP-tagged protein and started making some colourful blots.
    Now is the Licor Odyssey any diff or better than GE ImageQuant? I’ll bet you $5 the GE rep was a better salesperson. Because really, that’s all that seems to define the machines these days.

  19. Nicolas Fanget says:

    I can’t remember what the official Nature policy is regarding gels because I’m not involved in peer-review here, but at SGM we reserved the option to ask for the full gel, and if you couldn’t provide it your chances of acceptance dwindled…

  20. Richard P. Grant says:

    Colorimetric? ECL? Hah. I cut my teeth on I-125 labelled secondaries… and (semi-)quantitive to boot.

Comments are closed.