The interlab study is an attempt by iGEM HQ to conduct a comparative analysis of the different methods employed by the varied and diverse teams internationally to arrive at useful data. A key problem in science is ascertaining absolute measurements; there is no point in one measuring the fluorescence of some given part, only to arrive at arbitrary units whose meaning to other scientists is near zero. Standards must be set, in order to embue our results with any useful meaning. The interlab study is a step towards that ultimate goal of blanket standardisation in the synthetic biological context.

It is suggested that a team conduct the interlab study as a preamble to the main event. However, we are the first example of an iGEM team at Warwick, and coalesced rather late in the day. Hence we decided just to dedicate three members of the team to pursuing it in parallel to our other endeavours, as a side quest. It was primarily undertaken by Dan Goss , Waqar Yousaf and Chelsey Tye , with support from advisers Will Rostain and Sian Davies .

The experimental timeline

The remit of the interlab study boils down to constructing and characterising, albeit minimally, three devices. They share a lot of similarities, and the objective is obviously not to create some wacky new form of life, but to measure well characterised and well understood parts in order to measure the measuring equipment, as it were.

Therefore, over a period of about 1 month, we engineered the three devices from the brief via transformation, miniprep, digestion, ligation, and all the protocols you would expect (more on that below). What follows is a timetable of our experimental work in the wet lab:

Date Protocols and measurements
05/08/2014 Transformed all parts from kit plates, including an RFP-producing control
06/08/2014 Isolated/inoculated one colony from each and and grew them up overnight
07/08/2014 Miniprepped the overnights to secure sufficient plasmid DNA for digest
08/08/2014 Restriction digested parts and linearised plasmid backbones for assembly
08/08/2014 Ligated digested parts to produce devices 2 and 3
08/08/2014 Transformed ligation products over weekend
11/08/2014 Inoculated colonies of ligation products
12/08/2014 Miniprepped ligation products to ascertain plasmid DNA of devices 2/3
12/08/2014 Gel electrophoresis assay of these devices, with positive results
... Hiatus period (focusing on other work!)
20/08/2014 Transformation of all devices from plasmid DNA for measurement
21/08/2014 Inoculated colonies of each device (three biological replicates for each)
22/08/2014 Refreshed cultures in M9 minimal media in the morning
22/08/2014 Measured optical density and fluorescence with plate reader overnight
25/08/2014 Collected data from plate reader, but gain was set to 100 so had to repeat
26/08/2014 Inoculated colonies of each device (three biological replicates for each)
27/08/2014 Refreshed cultures in M9 minimal media in the morning
27/08/2014 Measured optical density and fluorescence with plate reader overnight
28/08/2014 Collected data from plate reader and imported into Excel for analysis
29/08/2014 The end!

Protocols and methodology

Many of the protocols mentioned above which we used were harvested straight from the iGEM website, but we also used content from previous iGEM teams and instructions packaged with kits. The specific materials and procedures can be accessed through clicking the relevant hyperlink:

As can be seen from the above timeline, we used gel electrophoresis to assess the veracity of our ligation products. The below picture demonstrates one such result:

To acquire the measurements necessary, we used a microplate reader, rather than a flow cytometer or something similar. To do this, we used the Tecan Infinite F500 , seen below.

This machine is very flexible, supporting various different plate sizes (from 6- to 1536-well plates!) and various modes of measurement (supporting absorbance wavelength from 230-1000nm and excitation wavelengths from 230-900nm). It is configured in to take the readings we were after using the following setup:

  • 96 well plate (see diagram below)
  • Temperature: 37 degrees Celsius
  • Absorbance measurement wavelength: 600nm (OD)
  • Excitation wavelength: 465nm (GFP)
  • Emission wavelength: 530nm (GFP)
  • GFP gain: 35

We then ran the programme ‘GFPandOD_overnight’, for which one cycle runs like so:

  1. Orbital shaking (amplitude 2.5mm, frequency 246.7rpm) for 200s
  2. Wait for 20s
  3. Repeat shaking for 200s
  4. Wait for 2s
  5. Repeat shaking for 200s
  6. Take measurements (takes about four minutes)

This whole process takes 15 minutes per cycle, and would run 80 cycles, coming to 20 hours, unless interrupted. Generally the optical density (that is, the growth of cells) has maxed out earlier than that. I stopped my measurements after 14 hours because growth had reached a plateau.

In terms of the protocol for preparing the samples and the plate for the reader, the first step was to concoct some M9 minimal media in which to refresh inoculated overnights before putting them into the plate reader. For 200ml of M9, combine the ingredients given, paying attention to amounts, concentrations and preparation advice given, and make up with H2O:

  • 20μl, 1M CaCl2 (A)
  • 400μl, 1M MgSO4 (A)
  • 200μl, 10mM FeSO4¬ (F)
  • 40ml, 5x M9 salts (A)
  • 3.2ml, 50% glycerol (A)
  • 8ml, casamino acids 5% (A)
  • 20μl, 10mg/ml thiamine (F)
  • 2ml, 2mg/ml uracil (F)
  • 2ml 3mg/ml leucine (F)
  • 50μl, 8-10M pH 7.4 NaOH

NB. Here A = autoclaved and F = filter sterilised.

The final solution should be 7.4pH, so usually best to make up with water, and then add NaOH carefully. Now that you have M9, the protocol is quite simple:

  1. Inoculate three colonies (biological replicates) from each device overnight in 5ml LB broth, as given in the above ‘growing’ protocol
  2. We need to refresh the samples in M9. First transfer broth to falcon tubes and spin down
  3. Pipette away or decant the supernatant and resuspend the pellet in M9
  4. Prepare 1:50 concentration of sample in 1ml of M9
  5. Put refreshed samples in a shaking incubator at 37 degrees Celsius for six hours
  6. Retrieve samples, get a 96 well plate. Any given well should contain 10μl of sample solution and be made up to 200μl with M9. Prepare three wells per sample (technical replicates)
  7. In the top row, put 200μl of M9 in every well; these act as the blanks
  8. Where possible, avoid putting samples in wells along the boundary of the plate
  9. Into all unused wells, pipette 200μl of distilled water
  10. Put the plate into the machine and run the programme defined above

Hence if my devices are D1, D2 and D3, biological replicates denoted by A, B and C, and technical replicates denoted by #1, #2 and #3, then the table below describes exactly my 96 well plate:

The point of three technical replicates is that it is easy to spot anomalies – they are the odd ones out. With regards to optical density, most samples were ending up at around 1, so to eliminate any anomalous data I used a difference threshold of 20% of this. That is, if any member of a triplet of technical replicates was more than 0.2 away from both other values at end of play, that datum would be excluded. This test led to no exclusions.

Then I considered instead the GFP measurements, and applied a similar protocol, using a 20% difference threshold specific to the datum being considered. If any overflowing was registered, as was the case for many incarnations of device 1, I considered the last full row of data, before any overflowing took place. In this fashion, I excluded from my data set D3 A #3, D3 B #1 and D3 C #1. Excluding these values prior to taking averages allowed me to then consider, by comparison of the final averages for each set of three technical replicates (unless one had been excluded), whether the inclusion of any biological replicates should be disputed. This analysis led me to discard all data related to device 3 (D3) since none of its three final averages were compatible with each other by my 20% test. I therefore held off calculating the standard deviation of these data.

Any excluded data has a red background on the Excel databook , which can be found here .

In terms of controls, he M9 blanks provided a baseline of optical density of about 0.07. Anything over that was contributed by the cells in the sample. The blanks were not significantly different from one another, nor had any given blank changed significantly over time. This consistency and display of constant values proved the M9 to not be contaminated.

The two quantities we measures were the intensity of green light produced by the sample upon excitation by the plate reader (that is, green fluorescence), and the optical density (i.e. absorbance) of the sample.

Timewise, a full run with the above settings would be 20 hours, but I ran it only for 14 hours with good results. and costwise, this is mainly a question of capital outlay. A used Tecan machine alone sets you back $10,000. But no iGEM team would have to fork out for this. Their costs are primarily the M9 materials, several of which are commonplace in most labs or cheap otherwise, and the well plates. One 96-well microtitre plate can set you back $50 or so, so this is probably the most expensive cost associated with repeated plate reading.

There were practical limits to the possible measurements. Preparation of the samples to put in the machine takes a good 24 hours, and the machine itself needs to run for a minimum of 10 or 12 hours with this configuration. It can also only handle one plate at once, although of course you could have more wells (although generally this sends the preparation time soaring accordingly). So you can maximally test 96x5=480 samples a week, and it would take a big chunk of your time. Expertise with the software is also required.

Optical Density

Units: The optical density or absorbance is the logarithmic ratio of light incident on a material to light which penetrates the material. It therefore has no units.

Precision: As a logarithmic ratio, OD is always greater than zero. The Tecan Infinite F500 specification says the full range is 0-4. Using the output data, we can determine that the Tecan is precise to 6 significant figures, although this seems to stretch to 7 s.f. when OD is in excess of 1.

Accuracy: Containing as it does a monochromator rather than filters, the F500 does not require calibration.


Units: The fluorescence measurements are relative rather than absolute, without units. The blanks, containing no GFP, serve as a reference from which to consider all other measurements.

Precision: Again by observing the data gathered by the Tecan, it seems to be precise to 7 s.f, but across the whole range this time. All my data sat within the wide range 1400 – 65000.

Accuracy: As before, this machine does not require calibration.

Measurements and results

Please see the accompanying Excel spreadsheet for all the measurements proper. It includes annotations, and identifiers for each column of data that follow the same naming scheme as in the well plate diagram above. Also included is the .asc file and equivalent text file containing the data in its original format.

The directly measured quantities are fluorescence and OD. We derived various quantities from the data, including OD/time, fluorescence/time, and fluorescence/OD. Much of the data for this is in the ‘Derived Quantities’ worksheet contained within the Excel document. And the relevant graphs are given below for devices 1 and 2, as well as in the Excel document. The exclusion of device 3 is explained in Section II.