Friday 28 October 2016

Getting better and better

Every day and in every way
I'm getting better and better
Émile Coué de la Châtaigneraie
Week 2 of the F&F microbiology course and we've been making up media and pouring plates to characterise environmental isolates a) from the air with fall-out plates b) from the nasal epithelium.  We have 7 pairs of students but needed only 6 sorts of media, so one pair was looking a bit idle so I appointed them autoclave liaison officer ALO and quality control manager QCM respectively. As ALOs they learned by doing how to safely and efficiently use the fearsome pressure cooker that sterilises the gloop that goes into Petri dishes. As QCMs, I asked them to inspect the previous week's plates and count how many were contaminated.

They found 2 plates with colonies in a stack of 32 'sterile' plates. French for Petri dish is boite de Petri; there, not a lot of anglophones know that.  I broadcast that information to the whole class and suggested that they incorporate the statistic, written as a proportion or percentage, in their write-up. I spent an hour the following afternoon reading and marking the lab books and correcting every report of 6.25% - which was all the books that had made the calculation. It is wrong, despite what the calculator tells you because, with a sample of 30 you have no confidence in the accuracy of the final .25 and so you are wrong to report it. 6% is more reliable I wrote multiple times. One intuitive way to grasp this is to ask what is the percentage if we had one more, or one less, contaminated plate?  3/32 = 9%; 1/32 = 3%. So abut 6% is plenty accurate.

While I was in the lab stacking up and storing away the solidified plates, I decided to get some more illustrative data and weighed two stacks of Petri dishes:
Chapman Agar, 50.0, 47.9, 51.2, 46.1, 50.9, 59.4, 45.2, 43.9, 47.6, 55.1, 49.6, 52.5, 52.5, 44.7, 51.8, Blood Agar, 42.9, 41.8, 50.5, 48.3, 46, 53.1, 44.2, 47.6, 47.8, 47.5, 51.4, 48.4, 51.2,
I calculated the descriptive statistics on each dataset and shared that with the class. It is desirable that all the Petri dishes have the same volume because this a variable. It's not crucial. Another way of saying it is that it is more efficient: if you can get in the zone and pour 15 plates precisely the same then you can carry out this infrastructural task a bit quicker and with less stress. The range and standard deviation are measures of departure from the central tendency; and it is clear that the Chapman chaps were more erratic in their pouring [std.dev 4.3; range 15.5] than the Blood boys [std.dev 3.4; range 11.3]. I sent these data to the class as an MS-Excel attachment. I'm waiting to get some feedback that it is damn-fool stupid to report the weight of a Petri dish as 47.9g because the balance isn't that accurate and a tenth of a gram is only 6 grains of rice.  This cross-disciplinification is regrettably a rare event in most science, the mean and std.dev is Quantititative Methods QM we can't be having that in F&F. But actually it is part of an essential training in Mathsemantics [prevs] making maths work for your crap-detector in real life.

For the math-enabled it is useful to note that as the diameter of a Petri dish is 8.4 cm and the area is πr2 then you need to pour 6mm of agar into each plate to use up 500 ml of agar. Our students, at the end of a year, should be possible to pour 15 plates so that the weight is +/- 2g. I could do that back in the 70s.  Of course it's a dying art. First world researchers nowadays buy their Petri dishes ready poured - in batches of 1000; by a sterile robot; in a factory far far way. No variation, no contamination, no time consumed on the infrastructure - more efficient altogether and wholly disempowering.