I’m running all sorts of experiments and testing different things, so I’ll describe the types of things I’m looking at in my experiments – if you would like more detail about my actual experiments and the exact things I’m looking at (eg.how I extract the DNA), please either ask here or ask me in a chat session.
So overall, it depends on the experiment. Some of my experiments are looking to improve my knowledge of the ‘why’, that is why something is happening. An example of this has been some of my studies with hair. One of the problems with DNA in hair is that sometimes people can recover DNA and sometimes they can’t. So I’ve been looking at whether there is a reason for this – is it because the hair is a certain colour? Or that it has been dyed? Or is it because the person didn’t wash their hair often?
However, the major part of my research is coming up with new methods and evaluating new methods. Testing DNA for forensic identification has different requirements to research – as well as standing up to other scientists when you publish them, the methods used have to stand up to examination in court!
So in light of this, for the actual methods I’m developing to extract the DNA and analyse it,I’m looking for what is known as the three the ‘R’s of a method that is going to be used in forensic science: Robustness, Realiability and Reproducibility.
Robustness is showing that the method will be able to handle a wide variety of samples – so in the case of hair, it will work on as many hairs from as many different people as possible!
Reliability is showing the method is just that – reliable. The answer I’m getting has to be the right one – after all, you don’t want to analyse someone’s DNA and get a different result back, and then end up identifying the wrong person! To do this, I run experiments where I compare the results of new methods with current forensic methods, usually using a standard DNA sample that will work in both, and that I know what the answer should be. That way any errors or problems can be seen very clearly.
And finally is reproducibility. I should get the same answer when I run the sample multiple times, and that the method works regardless of who uses it. Being to get an experiment to work once and never again isn’t going to be suitable in a forensic lab! So I run experiments where I repeat the same thing several times on the same samples to demonstrate that even on a different day, I get the same result.
As with Janette, I am also looking at several different things. Each requires a different experiment – in some experiments you get the results you are looking for and in others you make steps towards the next experiment. For example, with the DNA extraction Janette has described – for her it is almost the aim of experiment, whereas I use it as a midpoint. I know I can get DNA out of the bacterial cells so my aim is not the DNA but using that to look for something, generally a bacterial gene called the 16S ribosomal subunit. This allows me to find out what bacteria are in the human tissue without having to grow them. It is extremely useful because remember less than 10% of microorganisms can be grown in the lab, the rest are too fastidious or fussy, so take a very long time or do not like the media we try to grow them on. In one of these experiments, depending on the technology that is used, you can get hundreds of thousands of pieces of information in a few hours! In other experiments like the microscopy, the process is more continuous and you can get results the same day without all of the discrete checkpoints. The catch with those experiments is that you don’t know that something hasn’t worked until the end. An example would be a DNA fragmentation assay. If bacteria invade human cells then those cells often get programmed to die. One of the steps in the dying process is that the DNA gets chopped up into tiny little pieces. I can stain the cells to see if the DNA is intact (in chromosomes) or chopped up. I don’t know until I do all of the staining, which can take 8 or so hours and then have a look on the microscope. If I did not bust the cell wall up enough, then the stain will not get into the cell and I will not be able to tell what shape (good or bad) the DNA is in. This is often a problem with new experiments and so scientists often spend a lot of time optimising – using slightly different concentrations of key chemicals or changing incubation times by several degrees or adjusting the pH so that the experiment works really well. Optimising can take several weeks depending on how complicated the method is and generally needs to be completed for each different cell type.
Unlike the Janette and Elise my experiments are done in a computer and not in a lab. I use computer models that simulate the ocean currents around Australia. I do not change these much except on how much the wind can affect the currents in the surface layer. We can vary this to see what effects this has on the critters we happen to be modeling at the time.
The models also have components that simulate how some fish and invertebrates (lobsters etc) behave. Things like moving up and down the water column in search for food, coming to the surface at night when it is dark, or escape predators when it is day by going deeper. Also can the species in question swim and if so in what directions and how fast. We can vary this as well, sometimes the behavior we believe to be the case is incorrect and changing it in the model helps us try to understand that better in the context of ocean currents.
As to what do I mainly look for? Well that depends on the question(s) being asked. Recently the question has been where do baby (pick your favorite commercially important sea critter) go after they hatch. So we use the models to try and see were they most like will end up after some period of time if they are released from some particular area. We then match this with habitat data see if these area are suitable for the little ones to grow up in. This helps the managers decide which areas of the coast are more important to protect than others.
We can also do this in reverse, run time backwards if you will, to see where a particular observed patch of baby fish may have come from. Models can be run backwards as well as forwards.
I am usually looking for chemicals that are formed when fuel starts to oxidise or degrade. When you store fuel for a long time, or if you expose it to very high temperatures (like in a fighter jet or ship) it starts to go ‘off’ in pretty much the same way that food does. When this happens, the fuel starts to develop gums and particles which are bad for engines and make them run inefficiently, or in really bad cases the engine can break down. That is not what you want when you are flying a jet plane at 1900 km/h!
Other things I look for are contaminants in fuel, and these can be water, dirt, fungus, bacteria, rust, tiny parts of metal and graphite bearings, rubber or plastic from seals and oil. Sometimes my lab gets fuel with very strange things in it – once we got a fuel with strawberry jam in it… and the worst one ever: fuel contaminated with sewage. YUK.
I’m running all sorts of experiments and testing different things, so I’ll describe the types of things I’m looking at in my experiments – if you would like more detail about my actual experiments and the exact things I’m looking at (eg.how I extract the DNA), please either ask here or ask me in a chat session.
So overall, it depends on the experiment. Some of my experiments are looking to improve my knowledge of the ‘why’, that is why something is happening. An example of this has been some of my studies with hair. One of the problems with DNA in hair is that sometimes people can recover DNA and sometimes they can’t. So I’ve been looking at whether there is a reason for this – is it because the hair is a certain colour? Or that it has been dyed? Or is it because the person didn’t wash their hair often?
However, the major part of my research is coming up with new methods and evaluating new methods. Testing DNA for forensic identification has different requirements to research – as well as standing up to other scientists when you publish them, the methods used have to stand up to examination in court!
So in light of this, for the actual methods I’m developing to extract the DNA and analyse it,I’m looking for what is known as the three the ‘R’s of a method that is going to be used in forensic science: Robustness, Realiability and Reproducibility.
Robustness is showing that the method will be able to handle a wide variety of samples – so in the case of hair, it will work on as many hairs from as many different people as possible!
Reliability is showing the method is just that – reliable. The answer I’m getting has to be the right one – after all, you don’t want to analyse someone’s DNA and get a different result back, and then end up identifying the wrong person! To do this, I run experiments where I compare the results of new methods with current forensic methods, usually using a standard DNA sample that will work in both, and that I know what the answer should be. That way any errors or problems can be seen very clearly.
And finally is reproducibility. I should get the same answer when I run the sample multiple times, and that the method works regardless of who uses it. Being to get an experiment to work once and never again isn’t going to be suitable in a forensic lab! So I run experiments where I repeat the same thing several times on the same samples to demonstrate that even on a different day, I get the same result.
Hope that answers your question!
0
As with Janette, I am also looking at several different things. Each requires a different experiment – in some experiments you get the results you are looking for and in others you make steps towards the next experiment. For example, with the DNA extraction Janette has described – for her it is almost the aim of experiment, whereas I use it as a midpoint. I know I can get DNA out of the bacterial cells so my aim is not the DNA but using that to look for something, generally a bacterial gene called the 16S ribosomal subunit. This allows me to find out what bacteria are in the human tissue without having to grow them. It is extremely useful because remember less than 10% of microorganisms can be grown in the lab, the rest are too fastidious or fussy, so take a very long time or do not like the media we try to grow them on. In one of these experiments, depending on the technology that is used, you can get hundreds of thousands of pieces of information in a few hours! In other experiments like the microscopy, the process is more continuous and you can get results the same day without all of the discrete checkpoints. The catch with those experiments is that you don’t know that something hasn’t worked until the end. An example would be a DNA fragmentation assay. If bacteria invade human cells then those cells often get programmed to die. One of the steps in the dying process is that the DNA gets chopped up into tiny little pieces. I can stain the cells to see if the DNA is intact (in chromosomes) or chopped up. I don’t know until I do all of the staining, which can take 8 or so hours and then have a look on the microscope. If I did not bust the cell wall up enough, then the stain will not get into the cell and I will not be able to tell what shape (good or bad) the DNA is in. This is often a problem with new experiments and so scientists often spend a lot of time optimising – using slightly different concentrations of key chemicals or changing incubation times by several degrees or adjusting the pH so that the experiment works really well. Optimising can take several weeks depending on how complicated the method is and generally needs to be completed for each different cell type.
0
Unlike the Janette and Elise my experiments are done in a computer and not in a lab. I use computer models that simulate the ocean currents around Australia. I do not change these much except on how much the wind can affect the currents in the surface layer. We can vary this to see what effects this has on the critters we happen to be modeling at the time.
The models also have components that simulate how some fish and invertebrates (lobsters etc) behave. Things like moving up and down the water column in search for food, coming to the surface at night when it is dark, or escape predators when it is day by going deeper. Also can the species in question swim and if so in what directions and how fast. We can vary this as well, sometimes the behavior we believe to be the case is incorrect and changing it in the model helps us try to understand that better in the context of ocean currents.
As to what do I mainly look for? Well that depends on the question(s) being asked. Recently the question has been where do baby (pick your favorite commercially important sea critter) go after they hatch. So we use the models to try and see were they most like will end up after some period of time if they are released from some particular area. We then match this with habitat data see if these area are suitable for the little ones to grow up in. This helps the managers decide which areas of the coast are more important to protect than others.
We can also do this in reverse, run time backwards if you will, to see where a particular observed patch of baby fish may have come from. Models can be run backwards as well as forwards.
0
I am usually looking for chemicals that are formed when fuel starts to oxidise or degrade. When you store fuel for a long time, or if you expose it to very high temperatures (like in a fighter jet or ship) it starts to go ‘off’ in pretty much the same way that food does. When this happens, the fuel starts to develop gums and particles which are bad for engines and make them run inefficiently, or in really bad cases the engine can break down. That is not what you want when you are flying a jet plane at 1900 km/h!
Other things I look for are contaminants in fuel, and these can be water, dirt, fungus, bacteria, rust, tiny parts of metal and graphite bearings, rubber or plastic from seals and oil. Sometimes my lab gets fuel with very strange things in it – once we got a fuel with strawberry jam in it… and the worst one ever: fuel contaminated with sewage. YUK.
0