Summary of our Research

Contents:

1. What is Toxoplasma?
2. Our research
3. Analysing images using Artificial Intelligence and Machine Learning
4. How HRMAn helps us
5. How classifications work in our Zooniverse workflows
6. The importance of researching Toxoplasma

1. What is Toxoplasma?

Toxoplasma gondii is a parasite which infects humans and other warm-blooded animals. It is the most prevalent human parasitic infection with 1/3 of the world infected. Toxoplasma can only sexually replicate in cats - this is significant because the form of the parasite that the cat sheds in its faeces is stable and infectious. Humans become infected if we accidentally ingest it through contact with soil (e.g. during gardening, or children playing in a sand box) or contaminated water or food (e.g. unwashed vegetables) or from undercooked meat of infected animals.

Most of the time, our bodies can fight off the infection and we only experience mild symptoms such as fever, or no symptoms at all. However, it can be a different story for individuals who are pregnant or have a compromised immune system, for example due to taking immunosuppressant medicine (e.g. after having an organ transplant) or having HIV/AIDS. For these people, their immune system finds it more difficult to fight Toxoplasma. This can lead to the dormant Toxoplasma infection becoming active again which can cause a frequently fatal type of meningitis or visual impairment. When infection occurs during pregnancy, Toxoplasma can sometimes infect the foetus and cause health problems in the baby once it's born, or lead to spontaneous abortion (miscarriage).

You may have also heard of Toxoplasma being talked about as a "mind control" parasite - Toxoplasma can make infected mice less scared of cat urine (which they are usually afraid of because it means a predator could be close), making them more likely to get eaten. This is particularly beneficial to the parasite because when a cat eats an infected mouse, the Toxoplasma can sexually reproduce to form contagious oocysts, mentioned above.

2. Our research

Our lab focuses on understanding how Toxoplasma is recognised and destroyed by the cells it infects. In particular, we focus on how defence proteins from the infected cell are attracted (or ‘recruited’) to the parasite when it is hiding inside the cell, camouflaged in a layer of molecules which it has stolen from the infected cell on its way in.

To visualise these proteins and where they can be found in the cell, we use microscopes. We grow a special type of fluorescent Toxoplasma which glows green under a certain wavelength of light. Then we grow different types of cells and infect them with our fluorescent Toxoplasma, and add antibodies which stick to the protein we want to look at (the protein of interest). These antibodies have fluorescent molecules attached to them, which means that when they are observed under certain wavelengths of light, they give off colour (fluorescence) which allows us to see where the protein is in relation to Toxoplasma. This is illustrated in the picture below.

The images produced by the microscope look like the ones below - you can see green parasites, some of which are surrounded by the host cell’s defence proteins in pink. The proteins gather around the parasite-containing vacuole (the compartment where it hides from the cell's immune system) and alert the rest of the cell to the presence of the invader. This allows the cell to launch a variety of attacks against it, which can kill the parasite or stop it from growing. Open the Classify page to have a go!

3. Analysing images using Artificial Intelligence and Machine Learning

Microscopes are very useful for studying Toxoplasma, however there are still problems we need to solve: it takes a really long time to manually analyse and count all the images by hand, as there can be thousands of images from just one experiment! To solve this problem Daniel, a post-doctoral researcher in our lab, developed an automated platform which combines artificial intelligence and machine learning methods as part of his PhD research. This platform is called HRMAn, which stands for Host Response to Microbe Analysis and it helps us to analyse a lot of images in a short period of time and in lots of detail.

However, to get more accurate results we need to train it – that’s where you come in! We would like you to help us create a dataset which we can use to train HRMAn to recognise cell responses to Toxoplasma more effectively than it currently does. We would like you to decide whether you see proteins in the infected cell attacking the parasite inside or not. This will be done by looking at an image and choosing 'Yes' if you see proteins attacking the parasite, or 'No' if you can't see a response

4. How HRMAn helps us

HRMAn works by automatically detecting Toxoplasma in microscope images of infected cells. It does this by analysing the properties of each image including the sizes and shapes across different parts of the image. It looks at the patterns of pixels and how they are distributed in the images it is given, and compares them to the patterns it has been taught during the training process. You can learn more about HRMAn here.

HRMAn is already up and running and is useful for recognising Toxoplasma inside infected cells. However, we need to improve the consensus of whether the cell is attacking Toxoplasma with defence proteins or not. In other words, the more people help us to teach HRMAn whether there is an attack on the parasite by the cell or not, the closer we will get to accurately analysing our data!

5. How classifications work in our Zooniverse workflows

We would like you to help us create a dataset which we can use to train HRMAn with the input of many people. Like this, HRMAn will be able to recognise cell responses to Toxoplasma effectively in a consensus fashion. To do this, we would like you to decide whether you see proteins in the infected cell attacking the Toxoplasma inside or not. This will be done by looking at an image and choosing 'Yes' if you see proteins attacking the parasite, or 'No/I can't tell’ if you can't see a response.

If you come across an image that you find particularly difficult to classify, take a look at our ‘Tricky images’ collection or re-read the Field Guide. If you’re still struggling, you can choose the ‘No/I can't tell’ option – the images will go through another layer of processing later on to split them into smaller, more accurate categories. Overall, please make your best guess and consult the Field Guide and image collections but don't worry if you're not sure.

Below is an image showing the context of the parasite, i.e. where it is in the cell.

At the top, you can see the whole cell. The yellow shape shows you where the parasite you are currently looking at has been found in the cell. At the bottom of the picture above, you can see the parasite that you need to classify

6. The importance of researching Toxoplasma

While complications from Toxoplasma aren’t cause for concern for most people with healthy immune systems, some aggressive Toxoplasma strains can lead to serious complications in otherwise healthy people, including eye infections and even blindness. Gaining a better understanding of how our cells behave when they are infected with Toxoplasma will help us learn more about what happens during infection, how we can treat vulnerable patients better, and understand why some Toxoplasma strains cause more severe symptoms than others.

By helping us improve the software we use to understand this disease, it will aid the work of researchers around the world who use HRMAn. In the long run, this may help improve the possible treatment and vaccine options available for Toxoplasma, and other disease-causing microorganisms which can live inside our cells including Chlamydia and Salmonella. HRMAn is open source, meaning anyone can download and use it. This means that your classifications will help us re-train HRMAn to improve not only our research but the research of scientists around the world!