Pictured above, a non-ideal funding cycle

You know that long and frustrating funding cycle I talked about a while back, well the wheels have fallen off mine causing both a horrible mixed metaphor and my contract to runout.

When I finished my PhD, the Department of Engineering Photonics very kindly offered to keep me on as a PostDocs. As I mentioned before, there was no specific funding or project for me to work on, but they seemed to want to keep me and they found some money to pay my salary. The idea was that they would fund me until I managed to get a grant to pay my own way. This is a fairly typical situation that many PostDocs find themselves in – bring in grants or look forward to spending a lot more time at home…

So my main focus of the last 6 months has been identifying and applying for grants. It’s not the most fun you get to have a researcher, but it is a necessary evil. To be honest, I quite enjoy writing grant applications as it is in essence, an exercise in communicating your work and ideas to an unknown audience – which as you may have noticed from this blog, is something I quite enjoy doing.

The problem with this exercise in preparing grants, is that I can’t just bang one out and send it off; they take time to prepare and even longer to authorise (not as long as some of my other funding ideas but still quite a while). This is especially true of the grants I was preparing as collaborations with other groups. It is estimated that any grant from conception to money (assuming it’s successful) takes around 9-12 months. Unfortunately, it turns out that the department only had money to keep me on for 6 months – which, as those of you who are good at maths may have already realised, is less than 9-12 months .

As I said; originally it was hoped that the department would keep me on for long enough for one or more of these grants to mature but due to a number of external factors the department cannot afford to do this. In fact, a number of other grants are finishing around the same time and several of my colleges will also be leaving the department in the coming months. Such is the nature of current research funding – there is very little room to keep people on without project grants.

So what next…?

Well for me, I have been applying like crazy for another post-Doc position at other Universities around the country. While I would have very much liked to get some of my projects funded, I am quite looking forward to working in a new group and learning a whole new branch of science. I am also hoping that by moving to a bigger university I can look to taking my PGCEP (lecturing qualification) and getting back to the supervisory stuff I used to do during my time in Industry. While I am waiting to hear back about those PostDocs positions, I don’t plan on being idle with my time and you can follow what I’ll be up to on my new blog – The Errant Scientist.

Open Optics has been fantastic to work on for the department; I’ve really enjoyed writing it and it’s been really good fun talking to people about it and getting feedback from a huge variety of individuals. From a department perspective, it’s also been great in getting our work more publicity – Jane’s review article had a big up-tick in downloads after we publicised it here and is current sitting at the number 2 most frequently downloaded paper in the journal. Steve will be looking after the blog without me and has promised to keep it as up-to-date as possible. He won’t be posting quite as regularly as me, but I am pleased that it is now seen as an important tool of the department.

TL;DR – My contract is up and I’m leaving – because working for free is not good at feeding a hungry family. I am applying to new PostDocs positions elsewhere but in the meantime I am working as a consultant again and you can read all about my current projects on errantscience.com

(TL;DR = too long, didn’t read)

Professor Tromberg’s original talk was to be focused on “Medical Imaging in Thick Tissues Using Diffuse Optics” – a field that is one of many focuses of his group’s work at UCIrvine. However, when he was discussing the current activities of our group on his way into the University, he decided that actually we would most likely enjoy a broader talk which focused on numerous applications of photonics. So about 2 hours before he was due to give the talk, he re-wrote the presentation (by combining a few others, I assume…) and made us a whole new one and a half hour presentation on “Engineering Optics – from bench top to bedside”. Below are my notes on this talk.

———————————————————————–

Professor Tromberg’s first degree is in chemistry but his career has ended up moving him more towards photonics and medical applications. In addition to his many other roles, he is a professor of surgery (non-practicing) and has very close ties to active medical groups. His group is part of the Beckman Foundation and consists of around 20 faculty and ~120 researchers. The Beckman foundation is a group aimed at promoting research in chemistry and the life sciences, with a focus on developing new and innovated instruments and materials, as well as supporting young researchers – you can read more here.

Currently, a number of medical imaging technologies work on the principle of bringing the patient to the device. This can be seen for CAT scans and MRIs – both of which have a very high initial capital cost, high running costs and require skilled and often dedicated operators. The ideal model for future systems is to move from this patient-to-technology model and develop devices that can be used in a wider range of settings at the point of care (e.g. at home or in local clinics). Current working examples of photonics being used to promote this move to point-of-care include simple things such as smart phone apps that are capable of accurately recording your pulse.

NOTE: I had heard of these heart rate apps before but I had assumed they were one of many scam/hoax apps on the phone and wouldn’t be any good. However, after Professor Tromberg’s talk we downloaded one and tried it out and found them to be very accurate. I strongly recommend that you download one and give it a go (links to the apps we used are: iPhone or Android).

Other examples of enabling technologies derived from photonics, range from the now very common LASIK to the more recent use of optical coherence tomography to study and understand conditions of the eye.

The next phase in the exploitation of photonics is further into the field of diagnostics and using photonics systems to see beneath the surface of the disease. There are a wide range of techniques that can achieve this, ranging from nanoscopy (high resolution but short penetration) to diffuse tomogrpahy (deeper penetration at the cost of resolution).

Reproduced from Professor Tromberg’s talk with permission

Non-Linear Optical Microscopy

One technique that Professor Tromberg has worked with is using Non-linear Optical Microscopy in a clinical setting. These microscopes use a number of optical effects to image a number of structures within the dermal layers – collagen has a clear second harmonic generation; cells (specifically the NADP and FAD within them) show a florence reaction; structural features tend to be strong scatterers; and finally, lipids can be imaged using a technique known as ‘Coherent anti-stokes ramen spectroscopy’ (CARS). Some of the most striking uses of this technology was in the imaging of suspected lesions within the skin of a patient. In the image below you can clearly see individual cells and the distinctive change in morphology in possible lesions.

Reproduced from Professor Tromberg’s talk with permission

These microscopes have now been commercialised and are used in hospitals across the US, and are finding widespread use in looking at a number of biologically relevant structures.

In Vivo Microendoscopy

Doppler OCT image of a mouse cortex. Reproduced from Professor Tromberg’s talk with permission

Also within the UCIrvine group, Professor Zhongpin Chen has been at the forefront of the use of a number of in vivo optical techniques which allow for the imaging of internal lumens (e.g. airways and blood vessels). The combination of OCT and ultrasound allow for the building up of a complex 3D picture of the walls around the vessel and in the case of blood vessels, the build up of plaque. This technique is allowing doctors, for the first time, to accurately image regions ahead of stent implant surgery and subsequent monitoring of possible re-growth around the stent.

This kind of OCT imaging can also be used to create 3D maps of blood vessels using doppler OCT. This technique relies on calculating the correlation between multiple images to show the moving areas within a depth of around 5mm.

In vivo macroscopic imaging

This area of photonics imaging can be considered the domain of wide field cameras. This kind of imaging is often works by analysing reflected light from the surface of the skin over a wide area using one or more specialist illumination methods. Spatial frequency domain imaging (SFDI) is one such technique that is being used by surgeons within the Beckman Institute and uses a similar method to the Microsoft Kinect system for the xbox 360 – it projects a series of lines in the IR range on to the subject and then alters the frequency of these lines. Using the reflected data, it is then possible to build up a map of the underlying blood vessel structures to a result with low resolution (~5mm). In addition to the structure it is also possible to derive the concentration of oxygenated haemoglobin which is highly significant for a number of clinical applications.

SFID imaging of blood vessels. Reproduced from Professor Tromberg’s talk with permission

Despite the relatively low resolution, this system is being used by surgeons to determine the blood oxygenation of tissues during surgery. If for example, a flap of skin that is being manipulated during surgery looses too much oxygen supply, it may cause complications in future recovery. By monitoring a live image of the oxygenation of that tissue, the surgeon can see what further action needs to be taken.

NOTE: Professor Tromberg showed an excellent movie of this being used mid-operation however, it didn’t transfer over correctly so I can’t include it in this post.

This kind of macroscopic imaging is also being used along with fluorescent tags in cancer surgery to identify the lymph nodes within the patient. Traditionally, this was done using blue dye injected into the patient and visually tracked by the surgeon. However, this process can be error prone and deeper lymph nodes can be missed. The use of macroscopic imaging techniques can improve this by showing the surgeon the movement of the dye through a number of possibly hidden structures.

Diffuse optics

Finally, Professor Tromberg talked about the use of optical techniques to examine  biologically relevant information from a large area with diffused light sources. An excellent example of this being put in to practice is the use of NIR blood oxygen monitoring before and during surgery to improve survival rates and to identify high risk patients. This technology has now also been expanded to monitor patient reaction to common anaesthetics and provide a wealth of new insight into how they can influence patent recovery.

Diffuse optics are also proving vital in breast cancer screening for patients with dense mammary tissue that is difficult to resolve in routine screening procedures. These techniques help separate the higher risk patients and prevent either un-neccassary biopsies or missed diagnosis. Also like the monitoring of blood oxygen levels these diffuse techniques can be used to provide constant monitoring to asses the patient’s state and reaction to chemotherapy medication providing new insight for the clinicians treating the tumour.

Summing up

Professor Tromberg’s concluding remarks were that the work his group has undertaken would not have been possible without the links to other fields which provide insight into application requirements. Having researchers that are trained in both photonics and healthcare, has been vital to developing novel techniques to solve long standing medical problems that wouldn’t otherwise be apparent to people specialising in only photonics. He again re-iterated this in answer to Dr James’ question at the end of the lecture, regarding how to manage cross-discipline projects – “Ideally you need a dedicated SWAT team that can find applications for techniques you work with. A mixed team is vital is developing those technologies further”.

Over the years, I have had significant experience of lab supply websites. My previous life at Mediwatch meant dealing with a lot of suppliers and I can say that with only a few exceptions, it is always a huge migraine like headache. To give you some idea of the type of site I’m talking about i’ve made a mock up of a generic lab supplies website (see below).

Note: this doesn’t include the inevitable pop-up asking if I’d like to try their new version of the site – which appeared on 5 out of the 8 sites I flicked through while researching this article.

Annotated with cynical sarcasm

If I am honest, most sites are actually pretty nice to look at (no spinning GIFs or yellow text on a blue background) and their owners are clearly trying to make them as easy to use as possible. However, there are 5 common annoyances that few seem to have solved.

1. Obligatory laboratory models – Who are these people dressing in lab coats to be photographed and why are they so interested in whatever coloured beaker they are holding?
2. Search bars – Like a fool, the first thing I do on a website is try out the internal search system. Sadly, I have been spoilt by the likes of Amazon and e-Bay where the search function will actually show me all the products on offer at that store. Lab suppliers have not caught on to this and frequently use search bars that are so broken they either show nothing or absolutely everything that might have some of the letters I used in the search…
3. Categories – Having given up on the idea of just searching for what I want, I then turn to my last resort – categories. My best guess for these, is that they are prepared by just one person; so what you end up with is their slightly eclectic groupings – many of which only make sense to them. My real world example of this is when I found a retort stand was under ‘Chemistry>Experimental Equipment’ but not under the broader ‘Lab Equipment’ available on the main page.
4. Stock levels – A sales person from a large lab equipment supply company once told me that they always have a minimum of 3 of any item on their website. They also explained that 50% of the time they don’t actually have the item but order it in from their suppliers on an ‘as needed’ basis to save on having too much warehouse stock.
5. Contact form – Why can’t I just e-mail companies?? Contact forms are so impersonal and you are never quite sure if anyone actually got them. [ED: we have a contact form...] But then again, they are very neat and tidy and clearly better.

Now having, by some miracle, found my item (or the nearest equivalent I’m prepared to settle for to avoid more searching…) this is where things have actually improved a lot. In the not too distant past, many sites would make you contact them to get a quote on a £5.00 part – apparently the prices sheet was a closely guarded secret and would not be revealed unless personally authorised by the dark god of sales. Mercifully, they have seen sense on this and now almost all sites will give you the prices up front and in some rare cases they are then even mad enough to ship the part!

I wonder if Amazon have a lab supplies section…

For anyone interested the dots were made using a Lego X-Y-Z plotter and the top row shows a reaction to 1µg IgG and the bottom row the reaction to 1ng IgG

However as nice as these pretty pictures are, the one thing that I didn’t have any access to was image analysis software so I had no way of saying how much of a dot these dots were. This is where ImageJ came in.

ImageJ is free software package that is to image analysis what real butter is to a piece of toast – vital and delicious. Authored by Wayne Rasband at the research services branch of the National Institute of Mental Health in Bethesda, it is an amazing suite of very useful image analysis tools that are totally free!

In the image I showed earlier, if for example, I wanted to know how bright those two dots are – then ImageJ can just simply show me the image intensity over an area by selecting a transect across the dot which ImageJ then plots.

Okay, so it’s good with images but the plot function could do with a little jazzing up!

So in no time at all, I can switch from messy photographs to raw data about the profile and intensity of my results. This kind of image analysis is vital for a whole range of the work I do and ImageJ is a fantastic workhorse. I’ve even managed to use ImageJ for more mundane tasks such as calibrating the monolayer troughs I’ve used in the past for coating experiments. During the experiments it is vital to know what area of monolayer is being held on the surface of the trough so it can be dynamically controlled. However, over time the motors can drift and the original calibration can shift so the area needs re-measuring. Previously, people had done this by flipping over the 5kg trough and playing with the little pot motors until they got a sensible value – not very practical. A much easier way was to use ImageJ to calculate the area for me – I took a photograph at the open and closed position of the trough (with a ruler on the trough for scale) and then ImageJ simply provided the answer to the selected area.

I cannot understand why they ever made these troughs sort of awkwardly circular…

Beyond these simple analysis tools, ImageJ is capable of an insane number of other cool things. For example, when showing people the result from the fluorescent photographs from before, one simple way to make it more visually accessible was to convert the 2D photographs into the smoothed 3D surface plots.

Single 1µg dot

Or if I wanted to show a surface within a 3D space, I can simply convert it a rotating 3D animation.

If you stare at it for 1000 rotations there is something wrong with you

One last thing that i’ve just discovered is that I can hook ImageJ into python so in the future I can have these animation auto generating along side all my data analysis!

Summary: I quite like ImageJ – it makes pretty pictures. I just wish that other authors would be as keen to open up their software and make it available to the community.

At its core, a lab book is just that – a book in which you record your lab activities. Historically, these are interesting but they are secondary to the discoveries their owners have made, as lab books were used as much or as little as the owner felt like. For example, Leonardo Da Vinci’s note book is impeccably detailed and annotated and full of beautiful drawings and doodles of inventions, whereas Isaac Newton’s is a wall of impenetrable text in hand writing that a drunken spider would be ashamed of. Interestingly, this wasn’t just a difference of Leonardo being artistically talented and using it in his lab notes; there is some evidence that Isaac Newton deliberately complicated his notes with riddles and confusing wording in order to hide the meaning from anyone that he deemed not intellectually worthy of its content. In some ways, this over complication is not far off how scientific publications work today, sadly. However, despite the difference in style, both Isaac and Leonardo used their lab books to record their findings/thoughts and they would later use these notes to present their work as books and periodicals. The notes themselves were personal in nature and not designed for sharing or disseminating their work to others.

As science changed, so did the lab books. Companies running teams of scientists quickly realised that lab books were critical to sharing research between related teams and for retaining knowledge when a scientist leaves to work somewhere else. If all scientist X’s lab notes are written in short hand that only they can understand then that is no use to the company that has just paid him for the research! Outside of the commercial world, this sharing lab notes is vital to ensuing that you retain the fine detailed knowledge of the experiments that is often lost when it is prepared for publication. Good lab book keeping can ensure that new scientists can learn more quickly from those more experienced and not be doomed to repeat the same mistakes (and any experienced scientists should have plenty of mistakes/failed experiments to learn from!).

So to help any aspiring scientist write up a good lab book, I thought it would be useful to list a few top tips for paper based lab books. I have not quite worked out how exactly some of these fit with using an electronic lab book but that is an on-going project and I’ll post a guide for electronic lab books just as soon as I work out the best way of doing it!

1. Always write in pen, never ever pencil. This is for two reasons, firstly pencil can be altered which will ruin any chance of using your lab book as evidence that you were developing a technology first, secondly you should just write in your lab book and not get too hung up of neatness or getting things wrong. If you make a calculation error that you need to correct – cross it out and put the right value in if required. Having the mistake visible will remind you next time not to make the same error and to double check figures that you have had a problem with in the past.
2. Don’t remove/white out anything. If you didn’t get it from tip 1. then I feel it needs repeating – ‘don’t remove stuff’. Even errors, cock-ups and experiments that cause unintended fireballs need writing up – just because you got something wrong is no reason not to make a note of it.
3. Write in chronological order. Okay, this is a bit of a no brainer but I have seen lab books that people have just randomly scrawled in and they quickly become incomprehensible to even the author.
4. Date everything. You may not need to re-visit some of your work until months or even years later and you will have forgotten when you did a particular experiment. This is particularly important for matching up computer data to a particular experiment.
5. Write out the aims and materials of an experiment before you start the experiment. This step is a pretty good habit, as thinking through your experiment aim beforehand is a very good way of double checking that you are actually running the right experiment. Writing the materials down in advance is similarly an important way of double checking you are all setup and ready to run. Personally, I also find that once I’ve run the experiment, I am much more interested in running the next one rather than writing up the thing that has just failed…
6. Use your brain when including data. Ideally all results should go in your lab book but if you are generating GBs of data then this is probably inappropriate. If you can’t put the data in your lab-book then make sure that at the very least, you paste in a summary graph or table.
7. Maintain an index. This saves sooooo much time later when you are trying to find that experiment you did, you know the one with the beaker and that solution.

There are a few more things you should/shouldn’t do but they vary a lot depending on your field and where you work. From my experience training up a few new scientists the above tips are the ares that people most commonly seem to skip or forget. In the interests of openness, I have included the scanned page of my lab book I posted for “show you lab books day” a while back. I have highlighted the ares where I’ve failed to heed my own advice.

You’ll be pleased to hear I gave myself detention for this

My free time this week has been spent dabbing a small child with calamine lotion after he declared on sunday morning that he “feels a bit itchy’. Normally when my free time/personal life is a bit hectic I try to find some room in my work schedule to put a post together however, I decided to pop into the lab on monday to do a quick experiment and have only emerged since to excitably wave results at my colleagues.

So no blog post…..but I don’t want anyone to feel short changed so instead please enjoy this picture of some poorly labeled data. I can’t explain why yet but this is very very exciting and the first image of something interesting.

Personally I find looking inside the project logo more interesting

Last week I had the rather depressing task of announcing the end of our recent attempts to start a crowdfunding project. As I explained, the project essentially ended because the perceived risk of the project was too great. However, despite this failure I am determined to provide as much information as possible so that anyone wanting to try this for themselves can learn from where we went wrong.

I have already put up the majority of details on how the crowdfunding project was to be structured here and here (I also made a 4min video asking for money but I’m quite happy for that to never find its way online), but one thing I haven’t really dealt with is how we came to construct the project the way we did and how we dealt with the key challenges. The challenges raised internally were essentially split in to two key categories – intellectual property and public relations – both of which had further implications.

Project IP - Being an open access project the first hurdle we had to get over was the perception that the project might possible involve giving away valuable intellectual property. While other Universities have started wide adoption of the inevitable rise of open access science, we are a little behind and the University is protective of any existing or future IP. Our solution to this was to create a sensor project that was inherently un-patentable. While our sensor is certainly novel (as far as I know) it is actually a combination of several existing technologies that are well known. This combination approach by no means makes the development easy, as this combination has never been attempted before, but it does make the technology hard to protect. We argued that by revealing the project there was no IP to loose so no reason not to make it all public. After a little back and forth with our legal department this solution was deemed acceptable.

Discovered IP - Along side the IP behind the sensor, there was some concern that during the project IP may be generated that would again be potentially valuable to the University. The concern raised here is essentially, that if I discover cold fusion (seemed unlikely but hey, anything’s possible in science…!) during the project, who would own the IP as the discovery would be open access and therefore prevent any future patent protection. My solution to this was that as the project was being publicly backed to be open access, any IP generated should be similarly open access. It doesn’t seem very fair to have the public all chip in and pay for the project, only for us run off with any valuable IP. Besides, many other funding routes require the University to sacrifice their rights to the IP generated, so this isn’t all that different. The only change is that, rather than someone else running off with our IP, everyone gets to have the IP that was generated – which seems is more in the spirit of the project. I am unsure if this was an acceptable answer to the problem (we never got a clear answer), but if it wasn’t then I’m not sure I would have been comfortable running the project anyway.

Bad PR (project launch) - On first pass, I had assumed that open access science funded by the public, for the public should get people pretty excited – or at the very least ambivalent. It wasn’t until I gave it a bit more thought that I realised that just the existence of this project may have detractors. Obviously the project never actually launched but I could imagine two arguments that had a realistic chance of having been used against us. Firstly, people may react badly to a publicly funded University asking for yet more money from the public. There is a public perception that the Universities are constantly demanding higher fees and wanting more money to research ‘pointless things’. I’m not even going to touch on the grossly un-fair claims that scientists engage in ‘pointless research’ (e.g. best way to butter toast) as it has been well covered by other bloggers and stems from not enough education about how research actually works (hmm if only there was some kind of project that might educate people better…) However, especially in this time of austerity, of Universities asking for more public money might come off as slightly… err, hard to swallow. The second is slightly more serious – what if it is used as a political tool? Funding in Universities is just a teeni-weenie bit contentious and it could be construed as a case where a University is not properly served by the existing public spending and now has to beg for money in order to do the very work that it should be doing. This could then be used by anyone with an axe to grind, as a good example of why the current government funding model is failing… blah blah blah… While both of these are unlikely, they were outcomes that the University was obviously keen to avoid. Honestly, I had no idea how to tackle these – I could understand why this might come up but in the main, the only reaction I have ever seen to open-access projects is positive and supportive. If there was this kind of negative reaction to what is a broadly very positive and balanced project involving the public and schools, then those voices would surely be in the minority.

Bad PR (during project) - Thinking more practically – what if six months into the project, it turns out that our design was way off, and it becomes clear that the project can’t progress much further? This question I would answer in two ways : firstly, so what if it fails – projects don’t work out all the time; if I only ever worked on things I was certain of, then I’d never get any work done or science would be very very dull. Not talking about failures is a major problem in science and to be honest, I would be very proud be one of the few researchers that feels comfortable publishing their results even if it didn’t go according to plan. Secondly, I have never worked on a project that has failed that spectacularly as there are always more avenues of research to look at and other solutions to problems. The project could have been delayed due to some unforeseen issue, but I couldn’t see it failing completely in the first 12 months, especially as we had already done some very promising prelim work. Again, it is a part of science that things change and adapt as the project progresses – showing people this process was an integral reason for doing an open access project in the first place.

So I hope some of this article gave you a little more insight in to what we were trying to do, and into the issues and questions that the project raised. In some cases, these questions seemed easily resolved but others were much harder and I’m not sure we were clear enough about how these were either issues that didn’t exist or were much smaller issues than people believed. But as I explained at the beginning of my post, I hope that by sharing this on the blog someone else can see where we went wrong and hopefully do better!

If you have any comments or suggestions on any of the points raised, please let me know in the comments section. Also before I put this project to bed I wanted to take a moment to clear something up that’s been bothering me. Is it “crowd funding” or “crowdfunding”, I’ve had to write it a lot recently and I genuinely can’t make up my mind.

There will be no cake, science has not been done

Our crowdfunding project is dead…

As many of you know, over the last 4-5 months I have been desperately trying to launch a large crowdfunding project to raise £100,000 in order to fund 1 year of research into an oil spill sensor system. The key feature of the project was that I would be able to be totally open and share the story of research with a wider audience of people with an interest in science, as it would be 100% funded by generous donators. Not only that, but I also wanted to promote feedback from those funding the research to help drive the direction and encourage better public engagement with science. Ambitious goals, but I really felt that crowdfunding a piece of research was the best way to achieve them, or at least partially achieve them – I’m not sure one project is going to change the way an entire nation perceives science…

But it is not to be.

When I first envisaged this project, I realised that I was going to be attempting something never before tried at a UK university and something larger than any previous crowd-funded research project (that I know of). Because of this, I realised that the project would almost certainly be contentious within the University and we would need to make sure it had a clear “Yes you can do this” from those in charge. As I covered in my previous post about the frustrations of this approval process, getting a clear un-ambigious answer has not been easy. I won’t bore you again with the decision-making structure of the University but suffice to say that this project has now been talked about on every level available and the unanimous reply has been “Hmm, interesting… you’ll need to get approval from X”. This reply has led to the final decision on the project running around in circles for the last 3 months with absolutely no sign of progress what so ever. The reason for this people passing reply is because this is a very new idea and very different from normal funding, and the University is keen to make sure they have considered every possible risk, prior to launch. Unfortunately for them, because it’s new the risks are mostly unknown, and it is very hard to risk-assess unknown risks. So while there is no clear reason not to do the project, as we have answered all the basic questions about the obvious risks, the unknown risks still possibly exist. The end result being lots of people passing the project around waiting for someone to either take on the potential risk and sign it off or find a way of de-risking unknown risks….which might be risky.

So given this 3 month cycle of motivational fun that has endeared me to no one (turns out people don’t like being constantly asked to make a decision they don’t want to make) I am going to have to admit defeat. I can’t keep working away at a project that is going flat out nowhere, with no prospect of any of the problems holding it up being solved. I do understand the need for caution and consideration by the University, and a number of people have been fantastic at helping me get this project going, but unfortunately it’s just not been enough to persuade those that make the decisions to let it happen.

I will post again in the next week with a bit more of a post-mortem on where we got to in the project and what I hope others can learn from our attempt to go ahead with this grand idea. I really hope that while I may not be running a crowdfunding project I can inspire/help someone else to try. One thing that has remained constant throughout my working towards this project has been the enthusiasm for this idea from almost everyone I meet. If I can’t run an open project then the least I can do is be open about why it didn’t work.

Executive summary: Very new funding idea + unknown risk = no project

[ED: I did ask the author to do a cartoon to go with this post. The 'drawing' below is what he gave me]

[ED]: We are upping his dosage of sedatives

I deal with a LOT of spectral data – it is not unusual for an experiment to produce 0.5 GB of data, all of which needs processing into some kind of meaningful results. This is the perennial problem with a lot of current science – computers are great at running automated experiments generating massive amounts of data but not as smart when it comes to spitting out the results “It looks like it worked Dave, now try it at 25 degrees”. Although to put my ‘problem’ in to some perspective – when running, CERN produces about 15 petabytes annually, which works out at the equivalent of one of my experiments every second. But still, they have a pretty big team, supercomputers and some clever crowdsourcing working on that where as I have me and a 2 year old laptop.

For my work, the key component of the spectral traces is the peaks. In the looping GIF below there is a typical spectral trace from one of my early experiments (a trial back in 2009 I think). The graph is plotting the intensity of light at the wavelengths being transmitted through a fibre while it is coated with ~15 layers of material.

The is a repeating GIF of the change in a spectrum during the coating of a fibre

As you can see, there are a great number of peaks in the graph, some disappear mid-experiment while others appear. There is a clear pattern in the graph as the experiment progresses but what is needed is a simple break down of that pattern. The simplest solution to this problem is to just simply pick a number of peaks, record the values as each layer of coating is applied and then plot them. This solution is okay for a quick look but there are ~50 peaks in that data – manually tracking all of them will be highly time consuming and in all likelihood, make you go blind from staring at graphs. The spectrum may respond differently at different wavelengths so randomly picking points is a poor approach to take.

The next step is to just have the computer find the peaks for you. Most data analysis software has some kind of peak tracking software and can easily cope with finding the peaks in 500 sequential scans, which would then produced 500 little lists of the ~50 peak positions. Which is only slightly more useful that the raw data but is still a long way from being any interest because the computer skips a step that you would have easily done without thinking if you did it yourself, tracking the peaks.

When you watch the GIF I showed at the beginning of the post, your eye would have naturally tracked the peaks as they move to the right. A computer isn’t that smart – all it can do is scan each spectra and say “yup I found 50 peaks in this one too!”. This can be perfectly acceptable as if the number of peaks never changes then peak 6 in one spectra is likely to be the same as peak 6 in the next, so you can simply plot peak 6 in every one of the 500 spectra and hope that the peaks always correspond. Although this pretty much never happens as data is inherently noisy, complicated and generally a pain. Below is a little example of what more commonly happens when attempting to track peaks in this way.

Peaks can appear and disappear randomly so this shifting can happen several times in one set of data

So the solution to this is that you need to provide the computer with a few ‘smart’ ways of distinguishing what peak data needs linking to what peak. While these methods are very unlikely to be better than a human’s natural ability to look for patterns, a computerised method as the added advantage of being impartial and therefore less likely to confer its own conclusions. So to do this I wrote SIR, otherwise known as the Spectrum Interrogation Routine which is absolutely not awkwardly named after an obscure cartoon character…

SIR is designed to track the movement of spectral peaks with a range of different ‘smart’ peak tracking methods. It was important to give SIR multiple methods as the movement of spectral peaks can vary greatly depending on the the experiment, so it was necessary that the program could cope with a large range of peak movements. All of the below tracking systems work by building up the data one scan at a time so the question it is constantly trying to answer is “where does this list of peaks fit in relation to the previous n scans”

• Default tracking – This is an upgrade on the simplistic idea of just listing the peaks. The method I wrote for this uses the three previous points to try and ‘predict’ the position of the the 4th. The diagram to the right makes this a little clearer. If the new peak value is within ~5% of the predicted position then it is assumed to be associated with that data, if not then the same algorithm is run on the next possible peak. If the new point doesn’t match to any existing peaks predicted position, the program calls it a new peak. This is a good starting method as it can cope with a wide range of peak movements, assuming a high sampling speed.
• Moving average – Moving averages are pretty common but this method uses it for tracking the position of the peak – by taking a moving average of the previous n values and comparing the new peak value to this value. If the peak is within a multiple of the STDEV then it is associated with that peak. If not, again the process is repeated with other nearby peaks until it either finds a match or a new peak is created. This method is a ‘fuzzy’ alternative to the default method and works best for widely separated peaks to allow for a large STDEV multiple. This is particularly good with ‘noisy data’.
• Linear or Polynomial fit – These are more complex tracking methods as they require some assumptions about the effect that you are expecting to see in the data. For this, the program will create a small model based on a certain number of previous points and then compare the position of the new peak value. This model system is highly noise-resistant and allows for tracking of peaks that might disappear for a number of spectra before re-appearing. However, these methods only work where your data actually fits the model. If your data isn’t linear or polynomial then you are out of luck.

On the off chance that there is anyone interested in using this code for their own spectral tracking, then the software is available below.

• SIR (Windows Installer) – This will allow you to run the program without any additional programs on any version of windows above XP (inclusive)
• SIR (source VIs) – This is a ZIP file containing all the VIs I used to make the program. All are free to use under the creative commons access.

Please let me know about any problems, bugs, feature requests or suggestions you think might enhance the program – either through the comments below or via the e-mail address listed in the help files. One of the biggest problems with the code is that I would like this program to be as usable as possible but I only had a relatively small sample of data sets to play with. So if you have a dataset that doesn’t work for whatever reason, then please feel free to send it to me and I’ll try to tweak the code and make it work!

Discoveries in science require two things, knowledge and inspiration. The first of these two can possibly be swapped with luck as there are several examples of people stumbling on to discoveries with little prior knowledge. Reading past papers and looking at all the previous work in a particular problem area helps focus down your attempts at finding a solution, ruling out  failed solutions and possibly hinting at missed solutions. However, all this research will still leave you with a nearly infinite number of possible answers 99% of which are likely to be total dead ends. It’s here that you need inspiration to give you some clue as to where to take your research next. Let me explain with a real world example from last year.

Problem: We were putting together a grant to design a range of new sensors that could be placed inside air systems to detect various gasses of interest. The grand idea behind the project was that if you could put, for example, pollution sensors into a building you could design a ventilation system than could smartly monitor the intake air and if there was a sudden release of toxic or hazardous gasses the system would automatically close the ventilation thus protecting those inside. Alternatively if there is a leak internally the building’s ventilation system could detect the problem and re-route the air supply venting it away from the occupants.  As previously discussed we specialised in fibre optic sensor systems which are ideal for this kind of large scale distributed sensing in ventilation systems. The only problem is that there is not a simple way to make these fibre optic systems sensitive to specific hazardous chemical components.

Solution: None found. After weeks of research I couldn’t find a single chemical or biological route that would selectively sense the toxic compounds we were interested in (polyaromatic hydrocarbons). Our only option was to start screening our vast library of potential’s sensitive chemicals for their use with sensor systems. This part of the project would have taken years with no guarantee of success, which is not a viable option for short term (2-3 year) research grants. We were stuck. That is until I had a very un-connected conversation with people on twitter which started after I read a great article on promoting yourself in science (note to adults: I had had some wine which may have helped, note to people under 18: alcohol is bad). Tweets are organised and sorted thanks to the wonderful site that is storify.

This last tweet makes more sense if you just remove the random “do it”, I think a Nike spam bot may have got to me.

Trust me this is hilarious if you are a biology nerd, I almost spat out my wine….

The above is a quite enjoyably random conversation about fly genetics and using hedgehogs as little waiters for cheese and pineapple parties (if they made a cartoon about that I’d watch it). Fun but hardly much help in developing a polyaromatic hydrocarbon optical sensor. Except that it made me think about a subject I hadn’t worked on or read much about since my degree, Drosophila files. Drosophila files were the go to example for 90% of my genetics and proteins courses at university, they don’t live very long, they breed better than rabbits and researchers get less attached to them than Guinea pigs, in my text books they were referred to as the ‘perfect subject’ for genetic research. As I re-read the wiki page linked by Lou it dawned on me that those little buggers have been studied endlessly and, like bees, they are known to have a pretty powerful olfactory system capable detecting rotten meat for miles around. So If they are so well typified I wondered if anyone had expired the proteins that are involved in this particular smelling pathway.

With some googling I came across this paper about a family of proteins I had never heard of before called Odorant Binding Proteins (OBP). These OBPs are low-molecular weight soluble proteins which have a poorly understood role in the highly active sensing of odours in insects. A little more research showed that these little proteins are not widely known about but almost every paper I read seemed to indicate that they have strong semi-selective binding affinities for a wide range of vapours. Eventually through further reference linking I came across this paper by a group in Italy. They had isolated a particular mutant OBP that appeared to have some sensitivity and selectivity to the very polyaromatic hydrocarbon targets I was trying to sense.

Who knows this might be a total dead end, the protein might not be any good for the application we are looking at but, every bright idea is worth pursuing if it has a reasonable chance. I’ve e-mailed the group that wrote that final paper and I hope that with a little more work we can see if this is a possible solution to our problem, and can be included in a future project.

I think the point I’m trying to make with the above story is that while my knowledge of sensors, proteins and hydrocarbons are critical in designing new sensor, these would be useless without a little inspiration to drive the research towards particular areas. People often ask me where my inspiration comes from for my various ideas and projects, the only answer I can think of is, everywhere.

Now if you’ll excuse me there is a song on the radio that’s got me thinking about another little sensing project…

P.S. If you are looking for inspiration I would strongly recommend reading the blogs of the nice people I was talking to on twitter. They work with / talk about some cool stuff.

@AnneOsterrieder – Plantcellbiology.com

@Protohedgehog – Green Tea and  Velociraptors

@JohnRHutchinson – What’s in John’s Freezer

@LouWoodley – SPOTON event