Archive for April, 2007

systems biology

Wednesday, April 25th, 2007

My impression of the current biology is that on one hand, it is deepening the understanding of the functional roles of individual genes and proteins, on the other hand, it really starts moving beyond to consider the structure and function of cellular signaling pathways. The latter is called systems biology, which integrating the necessary elements participated in particular biology processes within the full biological context in which they function. This approach makes sense to me.

The topic recently has been intensively discussed, for example,

The application of systems biology to drug discovery
Caroyln R Cho, Mark Labow, Mischa Reinhardt, Jan van Oostrum and Manuel C Peitsch
Current Opinion in Chemical Biology 2006, 10:294-302 [link]

There is a diagram in the above paper illustrating the process of systems biology, I re-draw it here:system biology flow chart.

Biology is simple

Tuesday, April 24th, 2007

I have heard many people, usually non-biologists saying that biology is simple, shallow, … But I don’t think mean that all the complex phenomena in living matters are just “fake” observation, it is the underlying principle that are very simple (in contrast to some theories in physics per se). In other word, the paradigm is that there are very straight ford underlying principles and the complex reality is just the production of many times of the simple application of the principles.

Reality = (f(x1, x2, …, xn))^N;

for example, f(x1, x2, … xn)= 0.1 x1 + 10.7 x2 + … x100 / 3.

The strongest argument for this assertion is that, as claimed by Uri Alon on Nature, “the cells evolved to survive, and not for scientist to understand.” I’m kinda like this idea, partially because of my physics background or some might say arrogance and ignorance of physicists. Uri Alon has been done pioneering work on transcriptional regulation networks in E.coli, in particular, he introduced the “network motifs”. The paper is here:

Simplicity in biology
Uri Alon Science 446:497 (2007) [link]

However, almost all my biologist friends claim that biology is very complex and very hard to understand. One more thing is that, some very elite physicists published papers or presented on informal or formal occasions with a striking title “Theory of Everything” (or TOE). I also want to remind you of that, in a latest biograph of Einstein ( I can’t remember exactly), the author said that Einstein never gave a unified theory for the universe. What a smart man.

A good book: In search of memory

Monday, April 23rd, 2007

I started reading this book at nights on bed. I must say that this is a very good book in terms of not only the personal history of Jews in WWII but also talent scientific pathways. I’m very impressed by Eric Kandel, the author. Definitely I’ll record some of my thinkings of this book along the reading. Here is the link of Amazon [link].

Olfactory and pattern-recognition

Wednesday, April 18th, 2007

I’ve been fascinated by the olfactory perception in brain. In particular, how does it work in this magical way?

The olfactory network dynamics and the algorithm of the information processing in brain are fascinate topics in neuroscience. The brain can separate many odors at many levels of resolution, which magically performing both gross classification and precise identification. Experiments seemingly show that odor perception is more likely to bind together than to accumulate and mix odor elements. Therefore, the olfactory system identifies odors through pattern-recognition process, which has been extensively studied in computer science discipline. Although the pattern-recognition of olfactory system is much more complex than that of computer science, the framework of the processing should be same—the input information (the binding of odors to their receptors) is processed through a dynamic network. On the other hand, because of the size and landscape of odor space, conventional computational patter-recognition approaches, such as, linear discrimination analysis, artificial neural networks are not suitable for the olfactory pattern-recognition.

Paper: Telomere and its relation to human aging and cancer

Thursday, April 12th, 2007

My understanding of cancer and human aging was strongly rocked after read papers about telomeres. Normal cells after removed from living body only could reach a limited number of replication in vitro, and then halter further proliferation. They are flat and extended shape and remain active metabolically without any more division. The phenomena that cells could pass through only a predetermined number of growth-and-division cycles are termed senescent.

It turned out that the proper maintenance of the telomere and the responses to certain cell-physiological stresses dominate the replicative potential. The proper maintenance implicates that there is a internal clocking mechanism enabling tracking the number of cellular division and halter any further division once the predetermined division number is reached. A phenotype of the proper maintenance is that the telomeres in senescent are shorter than in normal stage. Sometimes, even longer telomeres can also cause senescent. The cell-physiological stresses can be induced when cell suffer in vitro and possibly in vivo.

Anything to do with tumor?
Yes, senescence is actually an important tumor suppressor mechanism.

Is cancer the side effect of human aging?
I hope not. :-(

This posted is based on a review paper:

Telomeres: Cancer to Human Aging
Sheila A. Stewart and Robert A. Weinberg
Annu. Rev. Cell Dev. Biol. 2006, 22:531-557 [link]

Signaling pathways from a more realistic perspective: space

Wednesday, April 11th, 2007

Systems biology has been focused on characterizing the dynamics of biology systems, components of signaling pathways, and the response function of cellular signaling pathways. Somehow, I end up with a notion that once the components and state-flow of information in the pathways are established, anything else is just straight forward. The receptors on the extracellular surface recognize the signal molecules, and then convey some messages through other molecules (signals and/or proteins) inward to control target protein or directly regulate gene expression. But, really, how does it work from a “physics” view? We are more comfortable with signaling with lights, electrons, … And we are dealing with space and time all the time. There is no space and time in the biological signaling pathways, it seems everything “somehow” work out the result.

I was accidentally impressed by one paper.

Space in systems biology of signaling pathways – towards intracellular molecular crowding in silico
Kouichi Takahashi, Satya Nanda Vel Arjunan, Masaru Tomita [link]

We need and actually must pay attention on how space, molecular mobility, and signal transduction reactions fully implement the signaling. Especially, other biophysical properties, such as composition and orientation of both inside cells and between cells are also very important and need to addressed and their roles should be fully investigated. In the paper, the authors state that the total macromolecular density is 50-400 mg/ml, which is far higher than typical in vitro conditions (1-10mg/ml). Of course, the easiest answer would be do every simulation at the atomic scale, which is actually the most luxuries and hardly succeed even given the most powerful computers today. Tools which targeting the semicullar simulation is then the alternative and practical solution. For example, MCell has been shown work very well for some signaling pathways…

One Affymetrix tiling array analysis tool: MAT

Monday, April 9th, 2007

Affymetrix has developed oligonucleotide arrays to tile all the nonrepetitive genomic sequences, of course, NimbleGen and Agilent also have similar products. Although I am not very optimistic with Affymetrix’s strategies, i.e., 25mer and perfect matches and mismatches, we still could extract very insightful information out of sometime troublesome ChIP-chip data. Dr. X. Shirley Liu’s group built a tool to analyze Affymetrix tiling array, so far, it seems very promising.

Model-based analysis of tiling-arrays for ChIP-chip
W. Evan Johnson, Wei Li, Clifford A. Meyer, Raphael Gottardo, Jason S. Carroll, Myles Brown, and X. Shirley Liu
PNAS 103, 12457-12462 (2006) [link]

The algorithm was designed as following:

1, Introduce a equation of probe affinity, which usually makes computational guys comfortable, of course, multiple parameters were introduced, which is very practical from a bioinformatics perspective.
2, Use a simple least square to estimate the parameters in the probe affinity equation. This is done within each chip, which make the whole method very appealing because sometimes, there is no any replicated experiments.
3, Recalculate each probe intensity by using the equation and estimated parameters.
4, Standardize each probe on each array –> t values.
5, Use a sliding window with user defined length to make a MAT score for each probe, again, they proposed how to calculate the score from t values only. A detail should be noted that, within the sliding window, there are minimum requirements for probe numbers and gaps between probe coverages. Again, user could define their own minimum.
6, Each region with a p-value below certain threshold(user defined) will be generated, those regions are enriched ChIP regions identified by MAT.

The authors claim that their method is better than other methods, such as, HMM and TileMap.

It is a useful method especially when you need analyze noisy Affymetrix tiling array data and don’t have and want to make your own methods.

Tools to simulate the cellular processes

Monday, April 2nd, 2007

I have been fascinated by two tools simulating cellular progresses. Well, I should have said that one with explicit ability and the other with potential. They are MCell and Matlab. I’ll talk about Matlab later and am concetrating on one application of MCell, which has been published in 2005.

Division accuracy in a stochastic model of Min oscillations in Escherichia coli
Rex Kerr, Herbert Levine, Terrence Sejnowski, and Wouter-Jan Rappel
PNAS 103, 347-352, 2006 [link]

The observed biological fact is that E. coli, the rod-shaped bacterium reproduces in a characteristic way. The cell elongates along its long axis and duplicate its genomes, then divides symmetrically into two daughter cells. The dividing plane in wild-type E.coli is at 0.5 with 0.013 deviation of the distance along the long axis of the cell. As stated in the paper, there are numbers of approaching to explain the phenomena from deterministic and stochastic models. The authors introduce a stochastic model starting with the reaction-diffusion scheme.

It is known that in simulations, essentially there are three steps, the first step is to construct models, which includes choosing parameters, building targets and their inner and inter relations, and algorithm to simulate the event. The second step is initializing and iterating the simulation process by implement the algorithm. The last step is then to analyze the simulation results, nowadays, it means we need show the results in a very attractive visual way.

They identified the parameters–proteins involved in the cell division process, the key players are FtsZ protein which forms a ring and mechanically implement the division and determine the division site. In the meantime, the factor which inhibit the formation of the FtsZ ring is the presence of the nucleoids, which is the precursor of the formation of two daughter cells. The Min proteins are also actively involved, MinC inhibits the formation of the FtsZ ring while MinD functions as a recruiter of MinC. The concentration of MinD is higher at then ends of the cell than at the center, and MinD oscillates from end to end of the cell. MinE is required for the oscillation of MinD.

The algorithm here is a series of chemical reactions. The system is then consists of a rode-shape container, lots of proteins (5400 proteins), their relation is quantized by the chemical reactions. And the algorithm to simulate the dynamic process is MCell. Using this stochastic model, the authors were able to simulate the precise position of the cell division.

MCell could also be used to simulate other cellular dynamics, such as signal transduction, cytoskeletal motion, etc. Very impressive tool.