Speakers in Step conference

It has been a while since I came back from the Step conference in Brussels and I guess it is time to say something about some of the speakers I had the chance of listening to. Probably the most interesting ones from my highly subjective point of view were James Bassingthwaighte (University of Washington), Brian Goodwin (former Santa Fe Institute Faculty, now at Schumacher College) and Denis Noble (Oxford).

This Step conference was not meant to be about science per se so the talks were definitely not of a technical nature. James introduced the Physiome project which, as you might know, is about putting together all the current and future knowledge about the human phisiology with
the aim of improving health care. The ideal result would be a giant simulation of the human phisiology that could behave like a real whole organism. Such system would allow physicians and other researchers to test therapies quickly and without nasty side effects
and study ‘what if’ scenarios.What James thinks we need are:

* Training (No use of sophisticated systems if physicians don’t use them)
* Databasing
* Standards (Too many groups out there and no way to compare or integrate their work)
* Modelling archives (I got a nice model, where do I put it for other people to play with?)
* Modelling tools

All in all a nice and light introductory talk. Everything he mentioned is quite reasonable although I am not sure if it is realistic to expect any of these things happening in the short term. People so far seem to be happy happy to come with their own models and not much effort is done to see if the results of one model are consistent with the results of the model of a different group.

Next talk came from Brian Goodwin who, although use to be in the Santa Fe Institute is know a professor of ‘holistic science’ (which looks quite a scary name for a professorship). The theme of his talk? Computational biology: a clash of cultures. The part of the talk which I found more interesting was when he dealt with the ambiguity of languages. Human languages are ambiguous and the meaning of a sentence gets shaped as we speak. This seems to be a good analogy to understand the language of genes which is also ambiguous (which is nice if you want to evolve it). In his view both human and gene languages have the property that are the best compromise between the effort that the speaker has to make to convey a message and the
effort of the listener. This is an interesting idea although I guess that proving it might be quite complicated (note to my self, should take a look at what has been published about this).

The talk from Denis Noble was also interesting despite the fact that his major point was: I have a new book (“The music of life”) go and buy it (which I might do). He made a number of points:

1. There is no gene for function (no objections to that)
2. Transmission of information is not just one way (same here)
3. DNA is not the only transmitter of inheritance (heard that before)
4. Law of relativity in biology: there is no privileged level of causality. Message to Dawkins: the gene is not that important.
5. There is no genetic programme (message to Monod this time).
6. Actually there are no programmes at any level
7. …and that means not even at the brain level

Cancer and stem cells

Canadian and Italian scientists have just came with research that adds further strength to the idea that mutations to stem cells are the main driving force driving tumour growth and ultimately cancer. Stem cells are non differentiated cells that can replicate indefinitely. When a stem cell duplicates this can lead to two stems cells or to a stem cells and a differentiated cell. These differentiated cells can perform useful things such as become muscle cells, breast cells, epithelial cells, etc. As opposed to stem cells, these differentiated cells lose the capability of limitless replication. Every time a differentiated cell divides, the telomerase needed at the ends of the chromosomes gets shorter. Eventually there is not enough for replication and the cell undergoes apoptosis. That is one reason why many tissues have a pool of stem cells that keep producing differentiated while needed.

The researchers tried to find out how relevant stem cells are for cancer growth. They show than when performing animal experiments (much more convincing than in vitro), animals with injected colon stem cancer cells are more likely to develop cancer than those in which non-stem cancer cells are used.

It all sounds reasonable to me: one of the capabilities that tumour cells have to acquire for the tumour to become a cancer is limitless replicative potential. If you inject into an animal cells that already have that capability, that should make it easier for the cancer to appear. Also, it is known that some tumour cells, as they mutate, might revert to an undifferentiated state with stem-like behaviour. Therapies that specifically target stem-cell cancer cells should be the next step since stem cells amount to a small proportion of the cells in the body but seem to have such a great potential in cancer initiation.

Evolution on a chip

Via Nature research highlights I found an intersting article in PNAS about how a group of researchers in Princeton are studying evolution in silico…for real!

Normally, when theoretical biologists talk about biology in silico they are thinking of computer models of biology, but this time the in silico referes to silicon chips that have been used to create patched environments, each one representing a different microenvironment (the main difference between the patches being the availability of nutrients). In these patches they placed colonies of E. coli and let them grow. The bacteria were allowed to move from one patch to the next using narrow corridors.

Interestingly but maybe not surprisingly, the bacteria move towards more promising neighbouring patches and some times, adapt, genetically and physiologically to the environment. Asides from some interesting experiments, the guys have been kind enough to produce some mathematical model to study the evolution in silico as well as analysis of what is the evolution of bacterial density in a patch as nutrient availability gets depleted and competition gets tougher.

It is really interesting stuff but it seems that they need to complicate a little bit more the patches in order to get more adaptation to the environment and less motility to the greener grass.

The Step conference

I have just returned from my trip to Brussels for the Step conference in which a Physiome project was discussed. We had talks from some very nice speakers about which I will write in another post.

The Physiome project (or at least what I understood about it after being exposed to the idea for the very first time during this conference) is a highly ambitious project (and that is probably an understatement) whose aim is to integrate all the current and future knowledge about the human physiology. The idea is thus a multiscale modular framework in which all the models about the different parts of the human physiology could be integrated. Such a model would have a tremendous impact on our understanding of physiology, let alone the potential benefits for pharmaceutical companies. For all of you who have any experience doing modeling of biological processes I guess I don’t need to tell you how (let’s understate it once again) challenging this could be. In any case I am fine with any (extremly) difficult project as long as the intermediate steps are worth something.

In my opinion, the guys in the Step project should aim at something quite modest such as some system by which modelers can integrate just a few models together so different groups can check the consistency of their models and their assumptions. This process will probably take a long while but eventually most modellers will be used to think of their models not in isolation but as something that has to make sense in the context of all the models being developed elsewhere. There should be some infrastructure so the models can be shared between researchers and some protocols and interfaces between models at different scales or across the same scale (say molecular, cellular or tissue) so there can be integration.

One of the speakers mentioned that the keywords in this project are multiscale and modularity. I suggest taking a look at the field of software engineering in which different groups and companies work in different modules and at different levels of abstraction. The software produced is expected to work with other software modules. Of course the complexity to manage is different in the Physiome project but I still think it would be a good starting point.

Off to Brussels

I will be away for a few days. I will be attending the Euroconference organised by EuroPhysiome whose aim is to build a virtual physiological human.

The website of the conference is here.

Will report back at the end of next week