Protocol
Welcome to our Web App!
The function of our website can be divided into three components.
1、SIMULATION OF GROWTH METABOLISOM UNDER DIFFERENT CONDITIONS
The user can look up relevant reaction names and reaction IDs by metabolite, which serve as important parameters for other functions of the software, such as simulating growth, finding key enzyme and so on.
Using the simulation function, users can limit the upper limit of their uptake reactions by entering carbon source uptake reactions, nitrogen source uptake reactions, and target metabolic reactions as limiting conditions, while maximising the exchange reactions of target metabolic reactions such as acarbose.
After input, click the submit button to get the maximum reaction flux, which can help researchers observe each reaction flux when growing under certain experimental conditions.
After input, click the submit button to get the maximum reaction flux and the list of enzymes in the target metabolite metabolic pathway, which can help researchers observe each reaction flux when growing under certain experimental conditions.
2.KEY ENZYMES POSITIONING IN THE TARGET REACTION：
For finding the key enzymes, We calculated the FCC value of each enzyme in the acarbose metabolic pathway through simulation, where the FCC value represents the influence coefficient of the enzyme activity Kcat value on the target reaction flux, and the calculation formula is as follows:
The user enters the target response ID "EX acar" at "target", and then enters the name of Carbon source absorption reaction "EX Maltose (reversible)" at c_ source, Nitrogen source absorption reaction "EX NH3 (reversible)" in Nsource. After input, click submit to get a list: from the list, users can get the FCC's recommendations of key enzymes from high to low and their uniprot numbers in the acarbose metabolic pathway, and intuitively compare the differences in the changes of production and enzyme consumption.
According to the ID number of the protein, the user can query the relevant information of enzyme from Uniprot database. The amount of FCC represents the impact of enzyme on the target reaction. The larger the FCC value, the greater the impact of Kcat value of this protein on the reaction flux of acarbose metabolism. The returned reaction name indicates the chemical reaction involved by this enzyme in the metabolic pathway of the target metabolite, according to which users can find the substrate of enzyme. The change of yield product can directly see the change of acarbose yield and the change of enzyme dosage when Kcat value changes 0.1%.
3. FIND THE TARGET AND SCHEME FOR TRANSFORMATION OF TARGET ENZYMES:
In order to determine the specific mutation site of target protein, we provide a link to Hotspot Wizard tool, which enables users to enter the protein number to query sequence, model the homology of protein without finding structure query, then analyze the three-dimensional structure of protein. Finally, give the recommendation of mutation hotspots near the catalytic sites and the recommendation of mutant amino acid based on improving catalytic activity by compare the sequences of homologous protein.
A special feature we provide is that users can choose to add single mutation sites manually, and get a best retrofit plan out of all posibilities with the largest Kcat value based on our auto-screening mechanism, thus being freed of repetitive wet lab experimenting procedure.
Once obtaining a new protein sequence, users can re-enter it into the DLK network, predict Kcat value again, and observe the change of enzyme catalytic activity. Users can also choose to update corresponding Kcat value in model, rerun the flux simulation, and observe the changes of target phenotype, so as to provide reference for wet experiment.
WebAPP
Try our software!!!
Feedbacks
Once our modeling platform was initially built, we contacted senior students in our cooperating lab to get feedback on the trial. We are happy to have received practical suggestions for improvement.
The senior students gave feedback that For researchers studying this actinomycete, macroscopic understanding of cellular metabolic growth becomes more intuitive, daily use of basic functions becomes easier and faster, and the ability to give reference changes in modified proteins can guide wet experiments.
But when they tried our software for retrofit solution recommendation, they got a high number of possible solutions, and conducting wet experiment verification was actually a relatively large workload for them, and they wondered if improvements could be made.
PRESENT IMPROVEMENT
In order to build a more user-friendly platform interface, we made adjustments based on these feedback.
1.Hotspot wizard can give multiple mutation hotspots for a protein, and each hotspot actually contains multiple mutation directions, so the user is actually given multiple possible scenarios. So we added functionality to the software to pre-screen by permutation and return a best solution directly to the user.
2.In the DLK interface, in order to facilitate users to predict the Kcat values of wild and mutant proteins, we designed the function of automatically querying the sequence by inputting protease ID, as well as advanced options that can manually add mutation site and direction, so that our prediction network can not only fill in the missing protease parameters in ecGEM, but also verify the protein modification scheme given by Hotspot.
3.We invited some users not formiliar with our system to try our platform and we have revised our styem according to the feedbacks we received. Based on the feedback, among many revisions, we have added default values, examples in the input boxes and explanation about the meaning and format of the inputs.
FUTURE IMPROVEMENT
At present, our software only uses the GEM model of built-in Actinoplanes sp. SE50/110 as model organism. In the future, we could design a model that automatically build a high-quality DL ecGEM according to the GEM model of a certain organism input by users, and realize a series of prediction and verification functions.
Our software can also be linked with large metabonomics databases and enzyme databases, so that the metabolic reaction information and enzyme parameters can be updated in the GEM model of biology timely, so as to ensure that the information used for software modeling is more genuine and improve the reliability of prediction.
When using Hotspot Wizard to identify mutation hotspots, it highly depends on the accuracy of enzyme structure and the results of structure analysis, so we can introduce protein structure prediction tools such as Alphafold2 to make its structure analysis more accurate. In addition, the Hotspot Wizard only recognizes mutation sites near active sites such as substrate binding sites and molecular tunnels. However, through deep learning networks, important residues of inactive sites can be identified. Therefore, there is still available for improvement in recommended notion of software modification scheme.
GEM can observe phenotypic changes by regulating the amount of gene expression, but our software has not yet achieved this function. In the future, gene expression regulation tools can be added to the improvement. In addition to observing the changes in target reaction flux caused by the up and down regulation of key genes, we can also try to activate the originally dormant genes in creatures by the interaction between genes, which is expected to explore the optimization of high-yield strategies of strains and exploring new types of antibiotics.