"
Team:USTC-Software/Project/Preview
From 2013.igem.org
(minor fixes) |
|||
| Line 52: | Line 52: | ||
<div id="TT" style="margin-top:0px;"> | <div id="TT" style="margin-top:0px;"> | ||
<h2 align="justify">Regulation Prediction</h2> | <h2 align="justify">Regulation Prediction</h2> | ||
| - | <p align="justify">There will be a room for improvement in calculating regulation by alignment. We use global alignment in our software now, but actually in biological field, local alignment is also very important. As a result, combining global alignment with local alignment gives a better calculation. On the other hand, we use BLOSM50 to score the alignment but whether this scoring mechanism is suitable for calculating regulation needs to be verified. And using machine learning to optimize our scoring mechanism will be our future goal. Suitable is the best! It will be a part our future work | + | <p align="justify">There will be a room for improvement in calculating regulation by alignment. We use global alignment in our software now, but actually in biological field, local alignment is also very important. As a result, combining global alignment with local alignment gives a better calculation. On the other hand, we use BLOSM50 to score the alignment but whether this scoring mechanism is suitable for calculating regulation needs to be verified. And using machine learning to optimize our scoring mechanism will be our future goal. Suitable is the best! It will be a part our future work enhancing the efficiency of alignment. Better alignment method and backtrace model will make our software more excellent dealing with giant and complex network!</p> |
<p align="justify">Meanwhile, it is feasible to some extent that using alignment to predict the regulation, but concerning the other information about interaction such as operator, protein second structure would be more helpful. Focusing on the core of regulation is the key of analyzing!</p> | <p align="justify">Meanwhile, it is feasible to some extent that using alignment to predict the regulation, but concerning the other information about interaction such as operator, protein second structure would be more helpful. Focusing on the core of regulation is the key of analyzing!</p> | ||
| Line 64: | Line 64: | ||
<h2 align="justify">Simulation</h2> | <h2 align="justify">Simulation</h2> | ||
<p align="justify">Hill equation we used is universal in lots of field and as a result it will cause the low accuracy in network modeling. Finding an appropriate equation is really important. Chemical master equation needs a large amount of calculation, so finding a equilibrium point between accuracy and speed will be an important part of our future work.</p> | <p align="justify">Hill equation we used is universal in lots of field and as a result it will cause the low accuracy in network modeling. Finding an appropriate equation is really important. Chemical master equation needs a large amount of calculation, so finding a equilibrium point between accuracy and speed will be an important part of our future work.</p> | ||
| - | <p align="justify">There will be a lot of uncertainty in biological process, simply mathematical equation is hard to simulate the high coupling in biology. How to | + | <p align="justify">There will be a lot of uncertainty in biological process, simply mathematical equation is hard to simulate the high coupling in biology. How to decouple and enhance the accuracy of mathematical simulation will also be a part. In the same time, it will be a good attempt to add other ways of simulation.</p> |
| Line 73: | Line 73: | ||
<h2 align="justify">Converse Prediction Accuracy</h2> | <h2 align="justify">Converse Prediction Accuracy</h2> | ||
<p align="justify">PSO algorithm depends on the number of particles and precision, but it will increase the time of calculation, of course. We are trying to combine different optimal methods such as genetic algorithm, machine learning, annealing algorithm which could take a big part in finding best solution.</p> | <p align="justify">PSO algorithm depends on the number of particles and precision, but it will increase the time of calculation, of course. We are trying to combine different optimal methods such as genetic algorithm, machine learning, annealing algorithm which could take a big part in finding best solution.</p> | ||
| - | <p align="justify">For giving a better directivity, we are trying to pick out the really useful data of prediction to the users. And users could choose | + | <p align="justify">For giving a better directivity, we are trying to pick out the really useful data of prediction to the users. And users could choose what they need more efficiently.</p> |
</div> | </div> | ||
| Line 79: | Line 79: | ||
<div id="TI" class="vg"> | <div id="TI" class="vg"> | ||
<h2 align="justify">Visualization GUI</h2> | <h2 align="justify">Visualization GUI</h2> | ||
| - | <p align="justify">A | + | <p align="justify">A fantastically visualized UI is a key point of good software, more friendly control, more vivid data visualization, better network output. All of them will be a big part of our future work that improving the aesthetics and interactivity of GUI. </p> |
</div> | </div> | ||
Latest revision as of 14:45, 28 October 2013
Future Work
Regulation Prediction
There will be a room for improvement in calculating regulation by alignment. We use global alignment in our software now, but actually in biological field, local alignment is also very important. As a result, combining global alignment with local alignment gives a better calculation. On the other hand, we use BLOSM50 to score the alignment but whether this scoring mechanism is suitable for calculating regulation needs to be verified. And using machine learning to optimize our scoring mechanism will be our future goal. Suitable is the best! It will be a part our future work enhancing the efficiency of alignment. Better alignment method and backtrace model will make our software more excellent dealing with giant and complex network!
Meanwhile, it is feasible to some extent that using alignment to predict the regulation, but concerning the other information about interaction such as operator, protein second structure would be more helpful. Focusing on the core of regulation is the key of analyzing!
Simulation
Hill equation we used is universal in lots of field and as a result it will cause the low accuracy in network modeling. Finding an appropriate equation is really important. Chemical master equation needs a large amount of calculation, so finding a equilibrium point between accuracy and speed will be an important part of our future work.
There will be a lot of uncertainty in biological process, simply mathematical equation is hard to simulate the high coupling in biology. How to decouple and enhance the accuracy of mathematical simulation will also be a part. In the same time, it will be a good attempt to add other ways of simulation.
Converse Prediction Accuracy
PSO algorithm depends on the number of particles and precision, but it will increase the time of calculation, of course. We are trying to combine different optimal methods such as genetic algorithm, machine learning, annealing algorithm which could take a big part in finding best solution.
For giving a better directivity, we are trying to pick out the really useful data of prediction to the users. And users could choose what they need more efficiently.
Visualization GUI
A fantastically visualized UI is a key point of good software, more friendly control, more vivid data visualization, better network output. All of them will be a big part of our future work that improving the aesthetics and interactivity of GUI.
