Wednesday, May 21, 2008

Mind reading, big pictures and little details

The one thing about being a postdoc is that you’re no longer green behind the ears, and have developed somewhat of a decent bullshit detector. You’re also a little less in awe of your boss or most other independent investigators. And hopefully most postdocs become rather good at reading between their lines when discussing projects or experiments with their bosses. At least, when the boss says “this won’t work”, or “just go and do this”, you don’t just take their word for this.

Here’s my thumb rule for those statements. If your boss has this really nice but far fetched idea, and is really excited about it, he/she will say “these experiments are easy, you need to go for it. All you need to do is purify a couple of proteins, throw in some reaction mixtures and read your data”, you can almost be sure that the experiments will involve about two years of backbreaking work involving 16 hour workdays that go on for months, and plenty of labor in the 4 degree cold room. On the other hand, if you go up to your boss and discuss a proposal/idea you have come up with and the boss says “that’s way too hard, and I don’t think it’s going to work”, it means (a) the boss isn’t really interested in that idea but wants you to work on his/her crazy idea (see above) or (b) hasn’t really thought much about it since you are the one who has come up with the idea or (c) both.

One sometimes wishes for those early days as a rookie graduate student, in awe of the boss, naïve, gullible and completely willing to try the most insane or undoable experiments possible as a thesis project, only because the boss is such a famous scientist, he/she has to be right.

**********

On a related note, you can safely assume the thumb rule that any senior investigator who is exceptionally creative with big picture ideas (and who hasn’t done benchwork for 20 odd years) is going to propose the maximum number of “that’s easy” experiments that hang on a slender thread of a hypothesis (the kind that goes “if this were true, then this and this and this will be true, and so this story will be awesome”. Except that the very first “if this were true” is a big if). You’re also certain to be doomed to some years of crazy and hard experiments that are going to tell you that the first “if” is false.

***********

The phenomenal growth of molecular biology through the nineties and now the past ten years has made many biological experiments amazingly easier, and I’m very grateful for that. But I think it has done enormous damage to good, quantitative biochemistry (and biology in general).

The beauty about a lot of modern molecular biology is that it is carried out using kits made by companies, which the average trained monkey can execute. You don’t need to know most of how it works. Also, the reagents and kits have been made so good that you don’t even have to worry too much about being quantitative in your experiments. They’ll usually work (somewhat). You can throw in a little bit of an enzyme or lots of it, or a little bit of salt or lots of it, use approximate concentrations and approximate conditions and still the experiment works. Here’s a real example. Set up a PCR reaction with a Taq polymerase kit, good primers and a plasmid template. Now spit into that reaction tube, double the volume of the reaction with your saliva, and then start the reaction. I can pretty much guarantee that the reaction will still work.

What this oversimplification of molecular biology has done is it’s created (or is creating) a pretty large number of extremely sloppy biochemists. I see far too many undergraduates or graduate students who (a) don’t really understand the concept of molarity, normality, salt concentrations, pH, metal or buffer effects and (b) also think that just because they think it doesn’t matter, it really doesn’t matter. For example, they think that they can purify a protein on an ion-exchange column using either a 100 mM sodium chloride solution or a 200 mM sodium chloride solution and it won’t matter much. Or their buffer can be at a pH of 6 or 7 or 8 and their protein will show the same activity. And most of them don’t think pH meters need to be calibrated (and don’t know how to calibrate it anyway).

All of this, combined with the extreme confidence of today’s kids, makes for very interesting presentations in student seminars where they show rubbish experiments, and then authoritatively state “I think our hypothesis is wrong” followed by “this company that supplies us with X reagent sucks. My experiment didn’t work because I think the reagent has gone bad”.

If you’re going to make a statement like that, you need to back it up with data. But who needs data these days?

We need some more hardass investigators, who won’t be scared to crack the whip.

11 comments:

bongopondit said...

they think that they can purify a protein on an ion-exchange column using either a 100 mM sodium chloride solution or a 200 mM sodium chloride

That is quite intense science - right up there with a chemist (about to finish his PhD, no less) who asked if I could use potassium nitrate instead of potassium acetate for making a buffer (I was new in the lab and was looking for the chemical)!!!

Totally agree with you on how kits-dumbing-down-basic-techniques phenomenon (I was going to write a post on it sometime, so thanks for sparing me the effort :) ).

One observation though: even 'back in the days', I had noticed that biologists, esp molecular biologists, without training in chemistry (which is possible in the US) were never too concerned about the quantitative aspects. Usually there is nothing wrong in it, as most experiments work without careful quantitation.
But I remember my huge surprise when setting up my first restriction digestion experiment, being told b the post-doc 'ummm.....add one to three microlitres of the enzyme'!!!

Sunil said...

bongopondit...good to hear from you. And do go ahead and write your post on kits-dumbing-down, it is always fun to do a bit of bashing. And i would have thought a chemist should know better between acetate and nitrate.

As far as restriction enzymes go, 1-3 ul sounds about right :-)

Anonymous said...

Sunil,

A very different perspective-- it is hard or next to impossible to do very quantitative experiments in cognitive/sensory neuroscience--the questions are hard and usually very vague and the answers tell us about a few neurons that we cannot (generally) observe directly. So, there is a lot of quantitative stuff there-- like mathematical modeling, which I do a good deal. So, in neuroscience, the problem is the opposite-- one can design highly mathematical theories, but not have the experimental tools to answer the questions.
The stuff about advisors is very true-- I recently met a faculty candidate from MIT who spent 14 yrs as a grad+ postdoc working on a totally crazy mouse genetics and cognitive neuroscience project. However, it has now succeeded and he recently published 3 papers in Science. It is truly fundamental work. The PI is a Nobelist and internationally famous for being crazy.

Sunil said...

Anon.....I agree completely with you on cognitive/sensory neuroscience, and actually that is a major factor which annoys me. It is really one reason why it is such a challenging field. But that makes it doubly important to really think hard about the experimental design and the limits of interpretations of the experiments.

As far as the faculty candidate you met goes.....wow! But sometimes I wonder if it was all worth it. Linda Buck spent some 8-10 years as a postdoc with Axel, and though it worked out well in the end, I wonder what those 10 years were like.

jules said...

Hi,

I'm Jules and I work at bestrank.com, a company interested in blog
advertizing. I found your blog engaging and I'm contacting you to ask if
you are interested in blog post sponsorship.

If you are interested, kindly mail back for confirmation(jules@bestrank.com).Please include your blog's URL.
Looking forward to doing business with you.

Sincerely,

Jules Viernes
BestRank.com

Anonymous said...

Sunil,

The real question, I think, is what do you do when you know that the adviser's idea will take years of work to succeed and you do not have those years or when you know that your idea will work, but the PI is less interested? Either way, one is in for a rough time, because in the former, the PI thinks that the experiment is easy and should not take long (so the student/postdoc is not gifted) or in the latter, the PI thinks that you are unruly. What do you do then? This is something that happens pretty often in many labs. How do you deal with it, if there is a way to deal with it?
-

bhupendra.kahar said...

HI FRIEND
Nice blog.your blog shows your idea.

Sunil said...

anon.....I can only answer your comment with a sigh. It does happen quite a bit, and I don't think there's any one way to deal with it. But eventually the PI needs to be convinced, and some convince better than others.

satori said...

like ur blog.. interesting stuff :)
cheers

Bharat Rajawat said...

mind blowing

Deepak Srinivasan said...

its amazing to read your comments on science, research and molecular biology for various reasons.
one- nostalgia. I used to be a biologist and was working in the late 90s early 2000s, using molec. bio techniques...I dont do that anymore.
So just hearing the buzz words thrilled me.
two- the angst of the chemistry loyalist/biochemial researcher...
you sound incredibly like how i used to sound in my head. dont know if it was my intense love for all things intellectually chemistry, coupled with concern about research procedures that molecular biology was easily "easing out", and then to add to it, the poor understanding of how "chemical basics" play a role successful and long lasting impact to science, especially life/health sciences...
still smiling :)
reason three- the PI bit.
whoa! that was bang-on..especially the 20 some years not doing bench work...and hence being out of touch with reality...of how experiments work :D

totally a PI syndrome...
whee...im using PI after soooo long!
hehehe
excuse my excitement
Deepak
deep.srin@gmail.com