Network interference
12 important questions on Network interference
What is a metabolic network?
Edges = interactions
Network composed of metabolites and their biochemical interactions in an organism
What properties can be included in a network?
Layers of information
What is an adjacency matrix?
Symmetric matrix. No directionality
- Higher grades + faster learning
- Never study anything twice
- 100% sure, 100% understanding
What is the goal of metabolic netwerk interference at cellular level? And at higher levels?
- C: Search for link between metabolome data and underlying metabolic networks. (find method)
- H: metabolite association networks: more about correlations, distinguishing healthy from disease etc
What are two similarity measures?
- Correlation based (linear relationships)
- eg pearson correlation
- Entropy based (non linear relationships)
- eg mutual information
- usually for this, you need very many samples
- Often in MNI: correlations are used
Why does MNI use correlations?
- Systematic relationships between correlation networks and underlying biochemical network was shown to exist
- A fingerprint of a particular condition
- Succesfull applications in gene network interference using transcript measurements
But mostly practical (correlations vs eg entropy)
How are partical correlations calculated when you have many metabolites?
- Full partialization: correct C1, C2, for all metabolites
- extensive and a bit strong
- Partialization on one: correct C1 and C2 for one metabolite, choose the one that minimizes Rpar
- Partialization on two: correct C1, C2 for two metabolites that minimize Rpar
What is the overall scheme for partial correlation calculation?
- Repeated observations required (because you are calculating correlations)
- Repeats will not reproduce same data, even in same conditions
- high number of repeats required to for estimating and siginificance
How can in silico models be used to test algorithms?
- Use kinetic metabolic models to generate in silico metabolic data
- Induce small variations
- See if MNI works
How to mimic biological variation in a model?
- Eznymatic : add variation to enzyme speed
- intrinsic: add random factor
- environmental: also random factor (works less well, more of a pattern)
How can you quantify quality of MNI method?
- G-score
- can only be calculated if true network is known!
- G-score= sqrt(TNR * TPR)
What is the payoff with partial correlations?
The question on the page originate from the summary of the following study material:
- A unique study and practice tool
- Never study anything twice again
- Get the grades you hope for
- 100% sure, 100% understanding