After several months of hard work by all team members, we built the model and completed the whole project. And then, we successfully verified several ideas mentioned in some papers, which proves that our tool can works correctly as we thought. It turns that our project is useful. Biology students and some of the professors mentioned that it's hard for them to know what will happen precisely in a cell if they edit genes. They don't know whether adding or knocking out a gene will influence other reactions in the cell, thus reducing the production of the objective metabolite. Of course, they can cultivate certain strains to examine the production rate. But the whole process would be complicated and time-wasted. Our project can precisely predict which gene is unnecessary by in silico flux analysis.

If synthetic biologists want to increase the production of particular metabolites, here comes some questions: what types of genes should be knocked out? What kinds of genes should be overexpressed? If they knock out a gene, will it influence the growth and survival of the microorganism? Due to the different culture conditions and many uncertain factors, it's nearly impossible to predict the precise production of specific metabolites. However, we can precisely predict what genes will possibly influence the objective metabolite through in silico simulation. More surprisingly, our software can recommend genes the biologists to overexpress or knock out to achieve their goals. Through calculation, our tool will precisely minimize the range of genes that may have a positive effect on your goals. After calculation, we can present several clear tables to show the results. So it's likely for them to spend less time finding a proper gene, which will significantly reduce the workload of synthetic biologists.

Before the synthetic biologists begin their experiments, our software can simulate the process and precisely predict how the flux of all metabolites, especially coenzymes, will change. For example, if they know the flux of ATP will probably reduce significantly, they might improve the culture condition or change some promoters to achieve their goals.

Our project has many attractive features that distinguish it from others, and we have used many experiments data to validate each of them. We testify the reliability of the prediction through three perspectives:

The Escherichia coli MG1655 in silico metabolic genotype: its definition, characteristics, and capabilities, compared the predicted mutant growth characteristics from the gene deletion study to published experimental results with single mutants. Since one of the functions of our software is giving the gene knockout strategy that can enhance the production of some specific metabolites or the biomass, we tried to recurrent the calculation process to get similar results, which proved that our software did well in this kind of prediction. We imported the same model, E.coli MG1655, in our program. Then we just set the mode as default, namely, strategy to search for the maximum biomass and the environment it gave. When we deleted the gene individually, ForeSyn gave us answers immediately. 

Step 1. Search the gene ID in ForeSyn
Step 2. Delete the gene respectively in the e.coli core model. We can see the flux of objective reaction(biomass) is corresponding with what has been showed in the table in the paper. As is clearly shown, after some of the gene knockout, such as pfkAB and gapA, the biomass declined. We can quickly indicate that to optimize the biomass. We'd better think twice before operating with such genes. So it's time to review the paper and try to check if our result is right. When it deletes the genes that have just mentioned, the maximum biomass reduces, and gapA even fell to zero. We're pleased that such experimental results fit well with our in-silico results.

In Genetic engineering of Escherichia coli to enhance production of L-tryptophan, the biologists do some researches on how to improve the production of L-tryptophan. Through their experiments, they find out the deletion of several specific genes can increase the flux of L-tryptophan. They figured out that two pathways involving phosphotransacetylase-acetate kinase （Pta-AckA） and pyruvate oxidase(PoxB) contribute to acetate synthesis at the beginning of overflow metabolism, which will affect the production of L-tryptophan as expected.

When we want to optimize the objective reaction, we need to increase the fluxes of specific reactions or activities by directly overexpressing the corresponding genes. Also, we need to knock out some genes so that the metabolites in critical pathways won't be overused. 

Through the data we searched online, Metabolic engineering of Escherichia coli for direct production of 1, 4-butanediol, some papers mentioned they edit some essential genes and successfully construct a Thr-overproducing E.coli strain.We use our tool to simulate the same model, and it turns out we can precisely predict what genes should be removed and what genes should be overexpressed. 

The model used in the paper is E.coli MBEL979, which is a slightly modified network of iJR904. So we use iJR904 as our model because E.coli MBEL979 is not included in cobra. 

The paper shows that the deletion of lysA and metA will increase the production of Threonine because the expression of them will consume the necessary precursors of Theronine—Aspartate, and Homeserine.br

So we edit the iJR904 model by set THRS, a critical reaction to synthesize Theronine, as the maximized reaction. Through the comparison above, we can know that the flux of DAPDC and HSST both reduce from a specific number to zero after we set THRS as the optimized reaction. The flux changes to zero suggests that the reaction is not included in the best solution space, so we can knock it out to increase the production of Threonine.

What's more, we can see the flux of PPC is increased after we set the objective reaction as THRS. And it is corresponding with what is mentioned in the paper that overexpression of PPC can increase the production of Therorine, which is because it can directly produce Oxaloacetate, a precursor of Threonine. Therefore, it proves that it can precisely recommend which gene should be overexpressed.