If a farmer wants to understand the health and fertility of their soil, they could send a sample to a lab for chemical analysis. This produces useful data, including levels of: 1) nitrogen, 2) phosphorus, 3) potassium and 4) pH. These data points can help the farmer decide which crop to grow, what type of fertiliser to use and how much to apply. But it’s well-known that soil is much more than a simple reservoir of farming nutrients. Soil contains millions of species of bacteria, fungi and archaea that form the most diverse and intricate ecosystems on earth. We wanted to understand whether a biological analysis of soil might complement the way that farmers understand their soil. So we’ve spent the last 6 months bootstrapping a soil genetics lab and running a proof-of-concept experiment using Oxford Nanopore’s MinION sequencer. Here’s what we found out.

Before we continue, we should mention that this is a long post… about 5000 words. We’ve provided a handy table-of-contents below to help you navigate. However, for those that want the TLDR version, here it is: We learnt a lot from this project. We discovered that it is possible to run sophisticated scientific experiments without your own lab space. We found signs that genetic sequencing of soil microbiomes could be a complementary addition to chemical testing. And we found that nanopore sequencing is an enormous leap forward in terms of usability of genetic research tools, but there are still a lot of technical problems to solve before biological analysis of soil could become a part of everyday farming practice.

Contents:

Ways of learning about soil health

Over the past few years Science Practice has developed a strong interest in the measurement of soil health. This is primarily down to our work developing SoilCards: a paper-based soil diagnostic device that measures the key chemical indicators of soil nutrition: nitrogen (N), phosphorus (P), potassium (K) and pH.

When we got our hands on MinION — a genetic sequencer developed by Oxford Nanopore — we saw another opportunity to investigate soil health. Rather than measuring the chemical components of soil, could we use genetic sequencing to better understand its biological components? And, could this biological information tell us something about the quality and fertility of soil?

MinION and SoilCards

In addition to our interest in soil, we were also really keen to understand the usability of MinION. The ultimate aim of Oxford Nanopore is to allow anyone, anywhere to measure any living thing through its genetics. This meant that, as part of this experiment, we decided to also investigate how easy it was for us, a company without a lab, to run our own genetic sequencing experiments and analysis. We wrote about this particular side of the experiment in our ‘Genetic Sequencing with Distributed Labs’ blog post.

Time for metagenomics

MinION technology is based upon a biological structure called a nanopore. These are naturally occurring proteins that form pores through the surface of biological structures such as cell membranes. The key property of nanopores is that they allow some molecules, such as DNA, to pass through them. MinION works by embedding nanopores into a chip that can have an electric current passed through it.

MinION in our office

Our MinION in the Science Practice office.

When a DNA sample is added onto the surface of the chip (called a flow cell) and a current is applied, individual DNA molecules begin to move through the pores. As each DNA molecule moves through a pore, each different base (A, C, G, or T) produces a characteristic electrical signal which is sensed and recorded by the MinION.

This was not our first time using MinION. We have had two previous trial runs, one at Public Health England and another at the University of York – where we first tried sequencing DNA extracted from soil samples.

Soil is a good sample choice for a first genetic experiment for two reasons: it’s very easy to get hold of and there aren’t too many regulatory or safety concerns to complicate things. However, it did mean that instead of sequencing just one species, we would be attempting to sequence every microorganism present in a sample — something that is known as metagenomics. This approach allows you to gain an understanding of the diversity of species within a sample and, according to our hypothesis, an idea of the condition of the soil.

Hypothesis: more diversity correlates with soil health

Soil chemical testing: Getting our hands dirty

To get started, we required a variety of different soil types. We decided to collect ten samples from a variety of different locations across London — from public parks to private allotments, and from open sports fields to hidden wildlife areas:

  • Vegetable patch in Coram’s Fields
  • Vegetable patch in Camden
  • A flower pot (recently filled with compost)
  • Finsbury Park
  • Clissold Park
  • Shoreditch Park
  • Hampstead Heath
  • Lea Valley Park
  • Hackney Marshes
  • Victoria Park

Soil samples collected from the locations above

Soil samples before being sent for chemical analysis.

Soil samples collected from the locations above

Allotments in Camden: vegetable patches with cultivated and fertilised soil.

According to existing chemical data on geological maps of London, as well as our own qualitative observations, these samples were meant to represent a spectrum of different soil types. For example, some soils came from cultivated areas, some from marsh land, and others from much drier locations. However, to have a true picture of their chemical diversity, we sent each sample off to be professionally tested by Anglian Soil.

Here is the summary of the results we received from Anglian Soil (‘Site details’ are based on our field notes taken while collecting the samples):

Site code Sample site Acidity (pH) Phosphorus (P) mg/l Potassium (K) mg/l Nitrate mg/l Site details
A Coram's Fields Veg Patch 7.00 121 510 58 Site: vegetable patch. Cultivated. Fertilised.
B Camden Veg Patch 7.50 107 650 68 Site: vegetable patch. Cultivated. Fertilised.
C Flower Pot 7.05 101 442 115 Site: large garden flower pot with compost. Fertilised.
D Finsbury Park 4.15 69 180 58 Site: park area. Growth: holly trees and bushes. Water: no open source within 20 metres.
E Clissold Park 7.25 57 560 47 Site: from beneath decomposing logs. Water: stream within 10 metres.
F Shoreditch Park 7.45 149 420 43 Site: patch of bare land. Growth: grass. Water: no open source within 20 metres.
G Hampstead Heath 4.15 47 236 73 Site: top of Parliament Hill. Growth: long wild grasses. Water: no open source within 20 metres.
H Lea Valley Park 7.15 186 1299 101 Site: next to a marsh. Growth: woods. Water: marsh/bog within 10 metres.
I Hackney Marshes 7.55 68 316 63 Site: edge of sports pitches. Water: canal nearby.
J Victoria Park 4.60 173 253 138 Site: park area. Growth: dense shrubs/woods. Water: lake within 10 metres.

To better understand the differences and similarities between our samples, we visualised our results in a parallel coordinates graph using RAWGraphs. As can be seen below, some interesting clustering of samples emerge immediately.

There are two distinct groups based on their acidity levels: three samples with pH levels below the 5.0 value and seven with their pH levels equal or above 7.0. There are also three distinct groups forming based on the levels of phosphorus (P) and nitrate. Interestingly, the potassium (K) values seem to be quite different across all ten samples, with only one major cluster caught in bucket number 4.

Visual comparison of soil samples in four key chemical properties.

Visual comparison of soil samples according to four key chemical properties: pH, P, K, and N levels. Anglian Soils group similar values from chemical testing, which we have used to plot each sample. A = Coram's Fields Veg Patch, B = Camden Veg Patch, C = Flower Pot, D = Finsbury Park, E = Clissold Park, F = Shoreditch Park, G = Hampstead Heath, H = Lea Valley Park, I = Hackney Marshes, J = Victoria Park.

Barcoding: Getting the most out of our experiment

For our experiment we wanted to try and test as many samples as possible with our two flow cells (this is because there is a limit in the number of times a single flow cell can be used). Because of that, we decided to use a technique called barcoding, which allows the addition of more than one sample to the same flow cell.

DNA barcoding works by chemically adding known genetic sequences to the ends of all of the DNA molecules in a sample. Each sample receives a different barcode; this means that after the sequencing is done, it is possible to identify which piece of DNA came from which sample using bioinformatic software.

Barcoding forces a decision between increasing the number of samples versus the quantity of data for any one sample. This is because there is a limit to the amount of DNA that can be added to a flow cell, meaning that each new sample in effect ‘dilutes’ the other samples analysed. It is possible to add up to 12 different samples onto a flow cell, but we compromised at 5 per flow cell.

As far as we know, we were one of the first people to try using barcoding to analyse metagenomes with MinION. We wanted to do this to increase our sample number for the experiment and to increase the power of our proof of principle experiment to be informative. However, the cost of diluting samples for barcoded metagenomic analysis is that it may prevent the detection of rarer species in a sample.

As clear as mud: From soil sample to soil species

Oxford Nanopore provide a cloud based analysis service called Metrichor for users of MinION. This includes applications that can separate data with different DNA barcodes into different files, and another called What’s in My Pot (WIMP) that matches sequences to bacterial, fungal, viral or archaea in a database of genomes from different species.

Left: Weighing out soil samples before DNA extraction. Right: DNA sample in the BioHackspace Lab.

Preparing the sample for sequencing on our first flow cell was seemingly uneventful. At first, it appeared we had large amounts of data to analyse and each DNA read (individual molecule of DNA that passed through the flow cell and was read by the sensor) successfully separated into its appropriate barcode. However, when it came to using WIMP to identify which species each read came from, a disappointingly low percentage of the total could be matched.

There are a few potential reasons for the low percentage of matched reads in the data. The fragments of DNA we had were relatively small, which makes it more difficult for WIMP to identify their species of origin (WIMP does not attempt to match reads below 700 base pairs in length). The short fragments may have been a result of the DNA extraction process used for soil, which is relatively abrasive, or of the repeated freeze/thaw cycles, as we stored the DNA in between the different steps of the sequencing preparation process. The standard protocol for MinION does not require storing DNA, however, as our distributed lab was working in more than one location, it was necessary to pause the experiment at certain points.

For our second flow cell we went from DNA to sequencing in as short a period as possible, avoiding freezing the sample. Unfortunately, our sequencing run was interrupted overnight, resulting in a very low quantity of data.

Despite these setbacks, we were able to use WIMP to identify distinct species in four samples. This allowed us to begin to make observations of correlations between chemical and microbiological data.

Site code Sample site Number of species identified by WIMP
A Coram's Fields Veg Patch 190
B Camden Veg Patch 133
D Finsbury Park 106
I Hackney Marshes 106

Meeting point: Investigating our data

At this stage of our experiment we had at our disposal three types of data for analysis:

  1. Data about the chemical properties of each of the ten samples from soil chemical testing by Anglian Soil,
  2. A list of different species identified by WIMP from our MinION metagenomic sequencing,
  3. Observational data about soil location, condition, and treatment made at sample collection.

We decided to proceed with the analysis of only four samples (A, B, D, I) for which we had all three types of data available.

We started by trying to better understand the similarity between soil samples. To do this, we decided to use Principal Component Analysis (PCA). PCA is a statistical procedure developed in the early 20th century. It goes by different names in different fields, and is currently in the limelight of machine learning. One of its uses is in visualising multi-dimensional data. As we had four soil samples and six chemical features (magnesium, phosphorous, potassium etc.), PCA allowed us to visualise this 6-dimensional dataset in a 2-dimensional graph.

Principal Component Analysis graph

Principal Component Analysis biplot generated in R to map the position of four soil samples (A, B, D, and I) in relation to six chemical properties measured by Anglian Soil: acidity (pH), phosphorus (P), potassium (K), nitrate, magnesium (Mg), and conductivity.

We transformed the original features into principal components that efficiently accounted for as much of the variability in the data as possible. We then picked the first two principal components for plotting purposes; in this case, they accounted for 93% of data variability. The six original chemical features (vectors in red) are the loadings of each soil metric on the first two principal components (PC1 and PC2). The four chosen soil samples (labels in black) are the scores of each soil sample on the principal components.

On the biplot, soil samples that are close together correspond to observations that have similar soil chemistries. Based on the more important PC1 (x-axis), the soil samples from Finsbury Park (D) and the Camden Veg Patch (B) have the least in common. The most chemically similar samples are the two vegetable patches (A and B).

Although our power to make strong conclusions is limited, the data does point to some interesting correlations that would be worth pursuing and validating in further experiments. For example, based on the results of our metagenomic sequencing, the most biodiverse samples were the soil samples taken from vegetable patches (A and B). Biodiversity in soil microorganisms is an indicator of soil health, which also correlates with our chemical data showing good levels of nitrogen, phosphorus, potassium and other nutrients. A and B were also the only samples that were taken from soil that was cultivated.

Number of shared species between each pair of sites.

Species shared between each site pair (including those not exclusive to each pair). The number of species shared between given site pairs is labeled in blue and indicated by the thickness of the blue connecting lines.

Overall, shared species between pairs of samples also showed interesting possible correlations. We calculated the Sørensen–Dice coefficient for each pair of samples. The Sørensen–Dice coefficient is a simple statistical method to compare similarity between samples. Values are between 0 and 1, with 0 representing absolute difference and 1 representing absolute similarity (sample pairs in table below ordered from most similar to least similar).

Site code pair Sample site pair for comparison Sørensen–Dice coefficient
A – B Coram's Fields Veg Patch – Camden Veg Patch 0.146
A – I Coram's Fields Veg Patch – Hackney Marshes 0.111
A – D Coram's Fields Veg Patch – Finsbury Park 0.105
B – I Camden Veg Patch – Hackney Marshes 0.105
B – D Camden Veg Patch – Finsbury Park 0.078
D – I Finsbury Park – Hackney Marshes 0.071

This method showed us that the most similar samples in terms of shared species were the vegetable patches (A – B); this matched our expectations from the chemical data. The most dissimilar samples were Camden Veg Patch and Finsbury Park (B – D), and Finsbury Park and Hackney Marshes (D – I). Camden Veg Patch and Finsbury Park are the most chemically dissimilar samples, but we would not have predicted that Finsbury Park and Hackney Marshes would have such distinct biological makeups. These samples are broadly similar chemically, however they have a significant difference in their pH values, something that could impact the microbial species represented.

We are treating correlations observed with caution. There is an intriguing possibility that samples that are similar chemically are also similar in terms of their shared species, but we do not have sufficient evidence to know if this is consistently detectable using MinION.

Unfortunately, it is not clear if the types of species that are shared between samples have similar roles in relation to soil. This, combined with the fact that conducting chemical tests is simpler and cheaper than sequencing, make it unclear whether this kind of broad species diversity measure adds value over chemical testing.

Currently, the actions available to somebody attempting to respond to soil health data are not generally capable of responding to data as granular as microorganismal species makeup. In this respect, improving access to chemical testing through products such as SoilCards, which we are working on separately, is a lower hanging fruit in terms of likely impact.

The potential of sequencing

It is true that species data is significantly richer than chemical. Available databases store a great depth of information on different species, including details about metabolism, growth conditions (including temperature and pH), pathogenicity, and more. This information could allow certain species to act as bio-indicators of specific properties of soil.

For example, species with roles in the nitrogen cycle, plant or human pathogenicity, or those that metabolise soil environmental contaminants could all act as indicators. However, isolating and drawing meaningful conclusions from this data for non-experts is a long-winded process, reliant on databases having the sufficient information to understand an organism’s role in soil.

For instance, there were numerous strains of Variovirax Paradoxus identified in all samples. This type of bacterium is known for its ability to metabolise a wide variety of pollutants in soils, promoting plant growth in otherwise contaminated soil. This is consistent with the fact that all samples came from London locations, as soil in cities is likely to have a large number of pollutants.

Also of potential value is information about crop or human pathogens. For example Janthinobacterium Agaricidamnosum is a type of bacteria that causes rot in cultivated mushrooms. All samples barring the Finsbury Park one (D) showed evidence of this bacteria, indicating that (all other things being equal) Finsbury Park could be a good spot to begin farming Portobello Mushrooms. Although this is said slightly tongue-in-cheek, it is not difficult to imagine that sequencing could be used to identify causes of disease in crops and aid treatment decisions.

We were interested to investigate if we could segregate different soil samples based upon the numbers of species observed with involvement in the nitrogen cycle. Our data did not allow clear segregation as all samples barring Finsbury Park (5%) had ~8% of species known to have a role in the the nitrogen cycle. Nevertheless, this line of analysis could produce highly relevant information for soil management, especially if combined with a quantitative knowledge of abundance of each species.

These specific stories are risky to be confident in. Validating species as bio-indicators would be a significant task. It is however a concept that deserves further investigation, as it could allow for a much richer understanding of soil.

PCA analyis with species similarity between samples overlaid

All species identified by MinION metagenomic sequencing by site. Each line represents an individual species of bacteria, virus, or fungi. Numbers on the left indicate the number of species per each site configuration (e.g. there are 28 species that only sites A and B have in common). Two species discussed above are highlighted: Janthinobacterium Agaricidamnosum found in sites A, B and I, and Variovirax Paradoxus present in all four sites.

Reflections

It is not entirely novel to attempt to sequence genomes found in soil. What is (to our knowledge) more unusual is the use of MinION nanopore sequencing. This technique allows direct sequencing of a sample, without the amplification of certain highly variable areas of the genetic code. MinION makes it simple to look at all types of microorganisms from bacteria, through fungi and viruses, to invertebrates in the same experiment.

Our experience was that the vast majority of species identified were bacteria. It was not clear if MinION sequencing is an effective way to capture all types of microorganism at once, nor the extent to which this mixed data could provide useful insights. As with any observational experiment in bioinformatics, our findings can only be as complete as the databases used for making comparisons.

One advantage of MinION sequencing is the possibility of measuring the quantity of each species identified, rather than simply the presence or absence. This works by measuring the number of DNA reads per species and can add an extra level of useful detail for analysis. Unfortunately, this was not feasible in our experiment, as by barcoding several samples into a single sequencing run we reduced the reliability for quantification.

Barcoding samples while sequencing metagenomes is something that we are not aware of others attempting. While we can say that this does work, each additional sample you add reduces the ability to identify species. To optimise this approach, it would be necessary to trial sequencing samples with different amounts of additional samples to identify the point at which the number of species identified drops below an acceptable level.

Something we did notice was the possibility of using barcoding to detect sample cross-contamination. When analysing our data, it was possible to detect a handful of DNA reads (fewer than 10) from a soil sample that we had not intentionally put onto the flow cell. Although this level of contamination is not a significant worry, the fact that it was detectable was intriguing. In labs with contamination issues, barcoding could become a methodology for tracing the source.

The short DNA read lengths found in our samples is another aspect that would need optimising. The method to extract DNA found in soil is relatively abrasive and so may inherently damage DNA. Could the number of identified species be improved by using any of the available DNA repair protocols?

Using MinION and Metrichor

One of the aims of our project was to gauge the usability of MinION by anyone, anywhere. While it was possible to use MinION, it does have a way to go before this ambition is realised. Oxford Nanopore are most likely aware of this, particularly with announcements of VolTRAX, a device that automates all sample preparation following DNA extraction, and SmidgION (yes it’s really called that) which miniaturises MinION further and will use a smartphone as its computing power.

As an example of the limitations, the need for a qubit for precise measurement of DNA concentration at several points in the process meant that we needed additional equipment which was comparatively expensive. While not a problem in a professional lab, it did add a hurdle for us and would likely add one for anyone else trying to do this without their own lab.

In terms of analysis tools, WIMP is very impressive, but it’s not yet at a stage where it can be easily used for this kind of analysis. For example, it’s not currently possible to compare multiple samples at the same time within the app. For this, data must be downloaded into an alternative analysis tool. Unfortunately, when data is downloaded, it is not easy to separate reads based upon taxonomic level (Species, Class, Order, Phylum, etc.) The display of species identified is attractive, but becomes very difficult to read when there are multiple species as names can often overlap.

We were trying to understand the role that each species identified plays in the soil environment. WIMP enables species identification, but offers little in helping to describe possibly relevant activities of each species, such as growing conditions or metabolism of specific compounds. This functionality may be beyond the remit of WIMP, but would be an incredibly helpful addition.

DNA sample in the BioHackspace Lab

The user interface for MinION — MinKNOW.

This point extends to genetic databases of microorganisms. In general, they don’t appear to be set up to ask questions about a microorganisms’ utility or impact within a particular environment. Although data about growing conditions and roles in various nutrient cycles can be found, this generally is on a case by case basis. For the specific domain of soil health, an easily accessible list of bacteria with a role in the nitrogen cycle would be an excellent resource (if this database does exist, please let us know!). This would increase the power of sequencing analysis by allowing these bacteria to be highlighted when identified.

Overall, the cost of genetics sequencing is very high in comparison to the cost of chemical analysis. Of course, sequencing is a new technology and its price has dropped enormously in recent years. Still, for our use case of measuring soil health, it will be some time before this technology becomes economical. For this to happen, the value in the data would need to be much easier to access, as well as the overall cost of reagents, equipment and analysis time will need to be reduced.

Moreover, if sequencing equipment (such as MinION) and bioinformatics software (such as Metrichor or WIMP) are to bring bioinformatics capabilities to wider groups of non-expert users together with all the value derived from genomic insights, they need to become more user-friendly. If you are familiar with our earlier blog posts (about our work with Genomics PLC, EBI Summer Bioinformatics School, or launching Sequence Bundles), you will know that this challenge is particularly close to our own interests.

Some of these criticisms may seem quite harsh, however it is precisely because of how far Oxford Nanopore has come with the MinION and Metrichor, that we are attempting to ask serious questions about novel applications for genetic sequencing. At the beginning of the project we wondered if MinION could enable people to monitor the health of their own soil. While this is feasible, the cost, preparation and analysis demands are simply too great for this to be efficient at the moment. However, it would be plausible for service laboratories using nanopore technology to perform systematic testing on soil samples in the future, in the same way that chemical analysis is being done now.

Conclusions

Our experience using MinION has been a fascinating and tantalising journey. Along the way we have collaborated and spoken with many generous and helpful people who have shared their time and expertise with us. In particular James Chong and Anna Alessi at York University, Lauren Cowley from Harvard (previously Public Health England), Mike Cox at the National Heart and Lung Institute, Joe Parker at the Jodrell Laboratory, the members of the BioHackspace in London, Ulrike Kauscher from Imperial College London and the staff of the London Genome Centre at Queen Mary University have helped us enormously to learn and experiment.

DNA sample in the BioHackspace Lab

Our MinION running at the BioHackspace Lab in London.

As a proof of principle, there is too little data to know for sure what is the most appropriate use of sequencing in relation to soil health. However, the correlations we have observed suggest that it is likely that sequencing can be a useful data type for exploring soil. The data we collected tentatively points towards the identification of species with specific known roles in soil nutrient cycling or plant pathogenicity, as useful information for people looking to better understand their soil’s health. Diversity as an overall indicator of health could also be of use, though it seems to be a more general measure.

A significant question for next steps would be to better understand how this information can be actionable. There are a few companies such as Mammoth Microbes, that are developing bio-inoculants to improve soil health by introducing specific bacteria to soil. While this concept is in its early days, paired with sequencing data, this option could become more precise and potentially more effective.

In order for this sequence data to be robust, a significant number of experiments would have to be carried out to understand how best to optimise soil sampling methods. Making sure that a soil sample is representative of a larger area is a significant challenge that this proof of principle does not address.

Overall, we think it’s truly remarkable that it is now possible for a company that does not have a laboratory to design and run genetic sequencing experiments. This experience has inspired us to continue our explorations in genetics and genomics, to increase our efforts to design useful tools to understand genetic data, and to create products that build on this understanding.

Running our own scientific experiments is a really important part of our work at Science Practice. However this isn’t always straightforward, not least because we don’t have our own lab. To get around this problem, recently we’ve been using a system we’re calling our ‘distributed lab’ - taking advantage of the networks and spaces available in London to get our experiments done.

Our most recent experiment has been in genetics. We’ve been trying to find out if we can use a new sequencing technology, called MinION, to identify different microorganisms found in soil using their DNA. Our idea was to see if it’s possible to use this data to understand soil health, as well as testing if we could independently run our own sequencing experiments.

Our findings from the experiment are discussed in a separate blog post, here we focus on our experience and learning from setting up our first sequencing experiment.

Our experiment

Up until very recently our goal of independently running genetic sequencing experiments would not have been possible; the cost of equipment was simply too high. However the introduction of Oxford Nanopore’s genetic sequencing device MinION has changed that. This new tool reads genetic code in a very different way to other available sequencing technologies, reducing the preparation cost and time for samples as well as allowing the device to be an order of magnitude smaller than other pieces of tech, making it portable.

Oxford Nanopore’s ultimate company aim is:

To enable the analysis of any living thing, by any person, in any environment.

This is a concept we find enormously exciting, adding to our desire to test out how easy it is for companies and people like us to organise experiments without a dedicated lab space.

Getting started

Before we went off on our own, we decided to get some help from people with experience in using MinION. We contacted researchers at the University of York as well as Public Health England who generously agreed to let us run trials with them. While the MinION protocol is relatively straightforward, the opportunity to learn from people with expertise was incredibly valuable. We’ve written about our experiences in these trials here and here.

Our trial experiments at Public Health England (left) and York (right)

Going out on our own

After our two trial runs we felt we were ready to run the MinION protocol independently. However before we could do this we needed to find a space in which to do it.

Initially we considered buying equipment and consumables (such as pipettes and ultra pure water) to run the MinION in our offices. However after the trials, it became clear that while it would be possible to run MinION in an office, it would be too costly for a proof of principle experiment.

Finding Space

London BioHackspace

The London BioHackspace

After deciding not to use our office as a lab, we needed to find an alternative space in which to conduct our MinION Sequencing experiments. Previously we had success working in academic labs, and of course had already made contact with researchers who may be willing to let us continue our experiments. However for this project we wanted to understand the demands of running sequencing experiments independently, and so decided against going down this route.

A big factor in this choice was our discovery of the London BioHackspace. This is a community-run lab in London that provides access to basic experimental equipment for a small monthly fee. Anyone is able to use it, provided they are given an induction and adhere to safety rules. The lab is built on a model of sharing equipment and knowledge between members, and makes perfect sense for anyone who is looking to find proof of concept of a practical scientific idea.

While we had found a space, and the bulk of the equipment needed, there were a few pieces of specialised equipment and reagents that weren’t (yet!) available at the BioHackspace.

DNA extraction kit (left) and DNA attached to brown magnetic beads in solution.(right)

These missing components presented varying levels of difficulty for us to get hold of. Some things we were able to purchase such as DNA extraction kits to isolate DNA from soil, or DNA binding magnetic beads and a magnetic sample tube rack that enable separation of DNA from liquids.

One piece of equipment that was a particular hurdle was a Qubit fluorometer. This device is able to measure the concentration of DNA very precisely, which is essential information when setting up a MinION sequencing run, but is prohibitively expensive for single MinION experiments.

After some research we discovered ‘iLabs’. This is a system that allows people to book equipment in professional service labs. Ordinarily this is intended for use for people in academic labs, giving them the option of hiring equipment rather than buying outright. However, after discussing our project with the Genome Centre at Queen Mary University they were happy to let us book time on their Qubit.

The Genome Centre

The Genome Centre at Queen Mary University

Another important component of organising experiments is the ordering of reagents. Essentially, this requires setting up accounts with suppliers of biological experimental reagents so that they will allow you to purchase from them. In our experience none of the suppliers we attempted to order from had any issue with this, although we were asked to explain our intentions very clearly. The process was likely made simpler as we weren’t attempting to purchase anything that carries any significant health risk.

Overall, the process of running our first independent genetic sequencing experiment provided us with some useful lessons. While individually none of the organisational steps were particularly complex, they often took longer than we expected. For example, as we are not an academic institution, we weren’t able to fill out applications to buy reagents as suppliers expected. This ultimately wasn’t an issue, but it did take a couple of days of back and forth emails to resolve, pushing back the experiment.

What we did find is that people are often very generous with their time and expertise. By contacting people directly and explaining your goals and motivations, it is often possible to organise a way to get what you need for your experiment. While not every city has a community lab space like the BioHackspace, making contact with academic researchers at a local university can be a viable alternative.

MinION at BioHackspace

The MinION running at the BioHackspace

It is important to note that running experiments this way can be somewhat risky, as the learning curve when trialling new scientific experiments can be steep, with few safety nets for mistakes. It is also time consuming. For us the distance between our office, the Hackspace and the Genome Centre wasn’t excessive, but was definitely not the easiest way to work. In particular, pausing experiments to allow for sample transportation has to be well planned, to minimise impact on results.

Distributed Labs

The idea to enable anyone with an interest in conducting their own scientific experiments to pursue that interest is something that has gained popularity in recent years through the citizen science movement. However, citizen science commonly focuses on more everyday equipment such as wearable technology and smartphones. Experiments requiring a laboratory stocked with pipettes and centrifuges, are more unusual, placing different demands on the organisers.

The process of organising space and time to conduct MinION sequencing experiments has highlighted to us the amazing potential of using community spaces and hired time in service labs. This ‘distributed laboratory’ model has enabled us to conduct our own proof of concept experiments using cutting edge sequencing technology. This is something that we were not certain was possible before we began the project.

The opportunity to run our own genetic sequencing experiments, and gathering first hand experience of the whole process of sample collection to data analysis was incredibly exciting, informative, and ultimately rewarding.

Checklist for using Distributed Labs.

  • If you are interested in trialling something that you’ve not done before, try asking researchers with experience in that field. Not all have time to help, but many are enthusiastic to share their knowledge with outsiders.

  • Try and find out if there is a community lab near you, and become a member. If not, try contacting researchers interested in a similar topic. There will be a limit to the time and resources they can spare, so you may have to tweak your project to be inline with their interests.

  • Alternatively you could set up your own community lab. This is a big undertaking, but very valuable. For help and advice try here, or contacting the BioHackspace.

  • Service labs sometimes allow people to hire time on specialised equipment. Email them, explain what you want to do and you’ve got a good chance at access.

  • Provided they are not harmful, reagent suppliers are likely to allow you to order from them. For molecular biology we used Thermo Fisher, Beckman Coulter, New England Biolabs and Qiagen.

  • MinION and Oxford Nanopore have an active and knowledgeable customer support, as well as an engaged community of people using their devices. If you need help, ask!

If you would like to talk to us about our experiences in setting up a distributed lab, please get in touch!

We’re excited to announce the launch of The Frontier – the world’s first venture-focused challenge prize! The prize is aimed at bringing together scientists, engineers and industry experts from all over the world to solve specific technical challenges.

The Frontier is launched by Hello Tomorrow, a global non-profit supporting science driven innovation, and developed by Deep Science Ventures (DSV), a VC backing scientists. DSV have allocated up to £500k in available prize & investment funding to successful applicants.

The Good Problems team at Science Practice will be working together with DSV over the next month to define specific technical challenges within three of the six high level challenge areas – optimising edible protein production, re-engineering soil, and better baby nutrition.

The Frontier Challenges

All six of the Frontier Challenges.

Scientists, engineers and other technical background individuals from all over the world are encouraged to apply and work together on solving specific technical challenges in the 6 highlighted areas. Applicants will have 30 days to form teams online and work on developing solutions. 30 successful teams will receive online support and mentorship from DSV and 10 will get to pitch at the Hello Tomorrow Global Summit in Paris, 26-27 October 2017, in front of a top lineup of VCs and over 3,000 science influencers.

We’re really excited to be collaborating with DSV on the design of the Frontier challenges and we look forward to seeing the emerging ventures!

If you want to start your own business, have a STEM background and are intrigued by one (or more) of the Frontier challenges, make sure you sign up by the 30th June!

Science Practice is excited to announce that our project SoilCards has been awarded £20,000 from the Cambridge-Africa ALBORADA Research Fund. The fund was established in 2012 with a generous donation from The ALBORADA Trust UK to the Cambridge-Africa programme.

In partnership with the Cambridge-based National Institute of Agricultural Botany (NIAB) and the Kenya Agricultural and Livestock Research Organization (KALRO), the award will support user research in Kenya this year. Our team will introduce the SoilCards prototype to farmers, extension workers, soil-testing services and agri-business dealers.

caption

We’ll be researching a range of topics which will help us understand the farmers who will be using SoilCards and their environment, such as:

  • what are farmer’s attitudes to soil-testing?
  • what water supply is available for use with SoilCards?
  • what variety of fertilisers are available and affordable?
  • how could SoilCards be distributed?

Follow SoilCards on Twitter to stay up-to-date with news about our visit to Kenya.

In September last year a new issue of the 2+3D Design Magazine came out and inside it — we had written an article introducing the three Science Practice themes. 2+3D is the biggest design magazine in Poland. The article was titled ‘Wieloboje w projektowaniu’ which in English translates to ‘Combined Track and Field Events in Design’ — a reference to the very interdisciplinary nature of our work at Science Practice, where it is not uncommon to see designers and researchers working together with geneticists, developers, science writers, illustrators, agriculture experts, aeronautics engineers, filmmakers, policymakers, data scientists… In fact — each project we work on at Science Practice requires a unique blend of expertise.

The word spread fast and not long after the publication of the article I received an invitation from the organisers of World Usability Day WUD Silesia to talk about our work at Science Practice at their conference in December 2016 in Katowice, Poland. I always enjoy such invitations, as they are usually a nice opportunity to look back at projects from a different perspective and see how they resonate with people. And because the theme of the WUD Silesia conference was ‘Sustainable Development’, I immediately thought about our Good Problems theme.

In 2016 most of our work in the Good Problems theme was focused on problems in the humanitarian emergency sector (namely, the WASH and GBV projects). Humanitarian aid is often defined in contrast to development work. Humanitarian actions are usually seen as short-term interventions (days, weeks or months) focused on the needs of people directly affected by a crisis. While the need to integrate sustainable principles in humanitarian aid is part of a growing trend in the sector, I realised that we had another project in our portfolio that was strongly rooted in the sustainable development context – this was the Longitude Prize 2014.

Sustainability and the Longitude Prize

The Our Common Future report completed by the World Commission on Environment and Development in 1987 defines sustainable development as:

Sustainable development is development that meets the needs of the present without compromising the ability of future generations to meet their own needs.

Central to sustainable development are two concepts:

  • the concept of needs;

  • the concept of limitations.

As we developed prototype challenges for each of the six candidate topics for the Longitude Prize 2014 (Antibiotics, Dementia, Flight, Food, Paralysis, and Water), it became increasingly apparent that most of these problems stem from our inability to sustain a technology or a service that we already enjoy into the longer future. Simply speaking — our current situation is unsustainable and unless we come up with something really good, we won’t be able to enjoy those technologies and services for much longer.

The Antibiotics candidate provides an illustrative example. Here, the problem we are facing in terms of sustainable development can be phrased in this way:

How do we keep the benefits of using antibiotics today, and not risk future generations’ right to benefit from using antibiotics in the future as well?

With this phrasing the concepts of needs and limitations can now be clearly defined:

  • our need is to be able to use antibiotics effectively;

  • future generations’ need is to be able to use antibiotics effectively in the future;

  • one limitation is that the more liberally we use antibiotics, the less effective they become (due to evolving microbial resistance);

  • another limitation is that the development of new antibiotics becomes ever more difficult and expensive, as it takes ever less time for bacteria to develop resistance to a new drug.

Longitude Prize 2014: Antibiotics - infographic

Longitude Prize 2014: Antibiotics infographic. Key things you need to know about the importance of the problem of rising antimicrobial resistance. Presented data put the needs and the limitations into a quantifiable perspective.

In June 2014 BBC announced that Antibiotics had been selected by the British public to be the final topic of the Longitude Prize 2014. In conclusion of over 6 months of our research and design work for the Longitude Prize, we proposed that the Antibiotics prize should focus on improving antibiotic conservation through better stewardship of the existing antibiotic treatments. This in practice could be enabled by a new point-of-care diagnostic that should help clinicians target antibiotic treatments more effectively. The challenge statement in the Longitude Prize rule book reads:

The Longitude Prize will reward a competitor that can develop a transformative point–of–care diagnostic test that will conserve antibiotics for future generations and revolutionise the delivery of global healthcare. The test must be accurate, rapid, affordable, easy–to–use and available to anyone, anywhere in the world. It will identify when antibiotics are needed and, if they are, which ones to use.

The competition is still running.

Interestingly, other candidates that did not become the topic of the main Longitude Prize 2014 competition were also rooted in long-term sustainability context. The goal of the Dementia challenge was to address the challenge of caring for a growing ageing population while relying on finite human and financial resources. The Food challenge was focusing on the need to produce more food to feed a growing population on one hand, and on the limitations of agriculture’s growing environmental impact on the other. The Flight challenge was looking for a breakthrough that would balance our need for a rapid and convenient means of global transportation with the limitations posed by environmental impact of aviation’s CO2 emissions.