Linux, computers and cure for cancer

[briandc]

Hello,
my first post here.
I've had an idea that I'd like some professional feedback on, as I'm a computer (linux) user but not a programmer. My question is:

just as an application can be made to scan an entire computer for a virus, why couldn't a computer application be made that can be used to locate cancer cells within the human body?

Since cancer cells have a very unique "code," and the entire human genoma has been discovered, couldn't a computer with a probe be used to scan areas of the body for cancerous cells, and eliminate them without the need of drugs or surgery?

Any feedback much appreciated!

Brian

[JPC]

It's hard to explain why the answer to this is 'No' as I'm guessing you don't have much of a background in genetics so I'll try a computer based analogy.

Your anti virus software can identify and remove errant code but if a component in your PC, say a processor, started to physically expand and spread over the circuit board then software wouldn't help, you'd need to physically remove it.

The code on your PC is information as is the virus, the code in your cells is physical as is the tumour, you can't expect to treat them as the same thing.

[severin]

Cellular reprogramming

I don't know. At first glance I agree with the first response however, I believe there is some experimental medical trials where immune cells are taken out of the body, then exposed antigens from cancer cells and then put back in so the body can target the cancer cells. But you can't linux to do this.

[briandc]

It's hard to explain why the answer to this is 'No' as I'm guessing you don't have much of a background in genetics so I'll try a computer based analogy.

Your anti virus software can identify and remove errant code but if a component in your PC, say a processor, started to physically expand and spread over the circuit board then software wouldn't help, you'd need to physically remove it.

The code on your PC is information as is the virus, the code in your cells is physical as is the tumour, you can't expect to treat them as the same thing.

Hello,
thank you for taking the time to respond. I think I didn't explain my first post too well.

If a computer is programmed to "find" a certain "code," and then remove/quarentine it, then it should also be able to be programmed to find cells that have a specific "cancer code" and eliminate them. In fact, cancerous cells have a unique code that is different from the normal cells.
My father had a tumor treated by cryogenics. If a computer was programmed to recognize those specific cells, and a probe that freezes or burns cells on a very precise level followed the computer's directives, I don't see why it wouldn't work.

The same approach would work for defective cells in the bloodstream. The computer would be programmed to locate and eliminate only those cells that are passing through the bloodstream, while others would remain unharmed.

In other words, all that the computer needs to be programmed to do, is to find the cells that have the specific cancer code for that particular individual. Just like searching a database for a keyword.


I hope that's a bit clearer..

brian

[GenoMax]

Since we are talking about biological systems one can actually use a bio-molecule to do the "scanning". Antibodies tagged with a drug/agent can, in theory, selectively deliver a payload to a specific location/cell type (search for antibody guided drugs/chemotherapy).

In practice biological systems are complex so this approach so far has had limited success.

[briandc]

If the computer can be made to "see" on the cellular level, then it doesn't even require an genetic understanding, but only some characteristic of the cancer cell: shape, color, size, whatever.

The key is to "teach" the app/computer to "see" the sick cells. That is all.


brian

[dpryan]

If the computer can be made to "see" on the cellular level, then it doesn't even require an genetic understanding, but only some characteristic of the cancer cell: shape, color, size, whatever.

The key is to "teach" the app/computer to "see" the sick cells. That is all.


brian

Much of the difficulty is in the fact that every cancer is different, though there can be similarities that can be exploited. If you have a surface marker, then you could develop some sort of nano device to target that. Short of that, you'll either need to get into the cells or have something that just sits around and waits for cells to start dividing quickly or monitors energy usage or oxygenation or some other marker. I know that people are working on various nano-related things, though I don't think we're to the point where we can develop anything useful yet.

Another question becomes, what should the computer "see"? There are a lot of possible marker for it to look at, and if you pick the wrong one (or use the wrong algorithm) your treatment will simply kill the patient.

Also, your "that is all" is laughably simplistic. Most of us here are quite proficient at programming, so we can tell you that that will prove exceedingly complex. There are many people working on new cancer treatments, but drugs and surgery are going to be needed for quite a while still.

Cancer is hard, there is no magic bullet to its treatment.

[briandc]

Much of the difficulty is in the fact that every cancer is different, though there can be similarities that can be exploited. If you have a surface marker, then you could develop some sort of nano device to target that. Short of that, you'll either need to get into the cells or have something that just sits around and waits for cells to start dividing quickly or monitors energy usage or oxygenation or some other marker. I know that people are working on various nano-related things, though I don't think we're to the point where we can develop anything useful yet.

Another question becomes, what should the computer "see"? There are a lot of possible marker for it to look at, and if you pick the wrong one (or use the wrong algorithm) your treatment will simply kill the patient.

Also, your "that is all" is laughably simplistic. Most of us here are quite proficient at programming, so we can tell you that that will prove exceedingly complex. There are many people working on new cancer treatments, but drugs and surgery are going to be needed for quite a while still.

Cancer is hard, there is no magic bullet to its treatment.

In recent years I've known several close ones die of cancer. Yes, it's a challenge.

Still, let's consider technology today: you go to the grocery store, and everything has a bar code. We can't read them, but the computer does, and it never makes a mistake.

They say that cancer cells, although there are many types (breast, pancreatic, etc), they each have a specific set of characteristics.

A person with lung cancer, will have one specific type of defective cell. Once its characteristics have been identified, it wouldn't matter if has metasticized (sp?), since that same cell characteristic would be the same no matter where it goes in the body. The computer would be able to individuate the sick cells no matter whether they were in an organ or in the blood.

Brian

[dpryan]

That's not how it works. Lung cancer can vary greatly not only between individuals, but within the same individual and over time. The shared characteristics between cancers are relatively abstract, you're not going to target individual cells using those easily.

[JPC]

In the most simple terms you are saying "if you can detect the cells and kill them you would have a cure" then 'Yes' that is true but that does not take into account the complexity of the biological system. I play rugby on the weekends and I know if I can catch the ball and run down the other end of the pitch then I can score a try but it's not happening as often as I would like (as demonstrated by our league position).

Size color etc are not going to be useful for identifying a cancer cell, there are already > 300 different cell types in the body, what if your computer 'saw' a liver cell and thought it looked like a cancerous blood cell?....zap, no liver! So you do need to detect on a molecular level and even then you have to ask 'when is a cell a cancer cell?', how will your computer differentiate between a skin cells that are growing rapidly to heal a wound and skin cells that are in the early stages of oncogenesis?

There is not one unique code that identifies a cancer cell, even within a tumour there are different cell types, between people 2 mutations in different genes can cause the same disease.

[briandc]

What I've understood till now was that there is basically one cell that goes haywire and replicates itself, causing a mass. But even if there may be multiple defective cell types within one organism, it makes little difference, since it is sufficient to insert the cell type and the computer will find it. Like a needle in a haystack. That's what makes computers so amazing.

Brian

[briandc]

In the most simple terms you are saying "if you can detect the cells and kill them you would have a cure" then 'Yes' that is true but that does not take into account the complexity of the biological system. I play rugby on the weekends and I know if I can catch the ball and run down the other end of the pitch then I can score a try but it's not happening as often as I would like (as demonstrated by our league position).

Size color etc are not going to be useful for identifying a cancer cell, there are already > 300 different cell types in the body, what if your computer 'saw' a liver cell and thought it looked like a cancerous blood cell?....zap, no liver! So you do need to detect on a molecular level and even then you have to ask 'when is a cell a cancer cell?', how will your computer differentiate between a skin cells that are growing rapidly to heal a wound and skin cells that are in the early stages of oncogenesis?

There is not one unique code that identifies a cancer cell, even within a tumour there are different cell types, between people 2 mutations in different genes can cause the same disease.

I'm not able to say how to find cells before they become cancerous, although if a computer can be programmed to detect certain indicator warning flags, then you could do that too. I don't see why not.

The key is to get the computer to distinguish sick cells from healthy ones. Just like an antivirus looks for a particular string or extension.

Brian

[JPC]

You are vastly underestimating the complexity of the human body, there are 90 trillion cells in each of us each and they are in a constant state of flux, each cell has 3 billion bases of DNA, also replicating, sustaining damage and repairing themselves constantly and only one change in the DNA sequence might lead to cancer I'm afraid it's not like looking for a needle in a haystack it's like looking for a single grain of sand in the universe.

Your computer needs to determine the 3 billion bases of DNA sequence of 90 trillions cells without causing any damage to healthy tissue. For as single patient there would be more data than exists on the planet. With the very best equipment it takes a few days to determine the sequence of a human genome and it always involves destroying the cell to get the DNA out.

[stefanoberri]

"In theory" it could be possible, but there are many serious obstacles to overcome.

At the moment "reading" the genetic code of a cell is disruptive. You need to process the DNA in such a way that you can't "put it back" into a cell. Even if you could, it is unlikely you can read the whole DNA of a cell without upsetting that cell.

Even if you could, it would be disruptive to read all DNA in all cells without damaging surrounding tissues/organs. How do you get to every liver cell without damaging the liver? You would need microscopic machines that run with no energy and automatically find EVERY cell and sequence their DNA. Today, such machines are the size of a large printer at best.

Even if you could, there are about 10^13 cells in an adult human being. Each has 6*10^9 nucleotides to be read (if you make a ASCII file with that sequence, each cell would require a 6GB file). Even if you had 1000 of such microscopic machines, each reading the DNA of each cell in one second, it would take about 317 years to finish. Furthermore, today, the sequencing costs approx 5000$ per genome and takes at best a couple of days each. You do the math.

Even if you could, it is not quite uderstood what makes a cell a cancer cell. Not for most cancers. If you give scientists the sequence of any of your cells, it is unlikely that they will be sure that that particular cell does not have a cancer mutation. What scientists can tell you, is to spot some well defined cancer sequences (search for EGFR, BRAF, KRAS, TP53). This is, of the above, the easiest bit, I would say, but still very very hard.

Computers are amazing and I love Linux, but I doubt this will be the approach.

[briandc]

You are vastly underestimating the complexity of the human body, there are 90 trillion cells in each of us each and they are in a constant state of flux, each cell has 3 billion bases of DNA, also replicating, sustaining damage and repairing themselves constantly and only one change in the DNA sequence might lead to cancer I'm afraid it's not like looking for a needle in a haystack it's like looking for a single grain of sand in the universe.

Your computer needs to determine the 3 billion bases of DNA sequence of 90 trillions cells without causing any damage to healthy tissue. For as single patient there would be more data than exists on the planet. With the very best equipment it takes a few days to determine the sequence of a human genome and it always involves destroying the cell to get the DNA out.

Is it any big thing for a computer to sift through billions of files to find one or two that have a specific code? Of course it might take a few hours, but it's something within its capabilities.
If a doctor can recognize a tumor on a CAT scan, why can't a computer? If the computer can recognize the tumor, then it is probably more capable of identifying each affected cell. A doctor cannot do that.

Basically, the computer would do what the doctor already does, but with much greater precision.


Brian