Description
Background and Innovation
With the development of molecular biology and the rapid progress of artificial intelligence big data screening, biomolecular diagnosis has been widely applied in medical field.
BUCT-China aims to establish a sensitive general diagnosis system based on the DNA Molecular Computing, to quickly and straightforwardly determine the pathogenesis and guild subsequent medical treatment.
We designed DNA molecular probes based on polymerase-mediated strand displacement reactions to form artificial biochemical circuits with neural network structures, it can perform weighted summation of multi-dimensional biomarker input signals, thus output intuitive low-dimensional results for in vitro diagnosis and disease identification. To validate its feasibility and superiority, we applied this system to infectious disease pathogen diagnosis and lung adenocarcinoma diagnosis.
The introduction of polymerase reduces the error caused by leakage and improves the accuracy of the system. In addition, the reprogramming of the weighting process reduces the design and manufacturing costs of the weighted chain and the complexity of the reaction, making it possible for the molecular diagnosis of high-weight targets. At the same time, cell-free systems allow for micro-reactions that help make the whole device portable.
Compared with cell-based biosensors, it does not cause problems of biosafety, while the requirements of equipment and operators are low, it is also helpful to improve the accuracy of early diagnosis of disease in less developed areas, and to provide guidance for Drug Administration.
Our innovation comes from our experience in the last three years of pandemic. As young generations we are more sensitive to the impact of the pandemic that brings to our lives. As the future engineers and scientists in the biotechnology and pharmaceutical fields, we have noticed that the diagnosis procedure must be improved to face such serious public health issues.
- As the World Health Organization (WHO) claimed that covid-19 is no longer considered a “Public health emergency of international concern”. It marks an important step towards ending the covid-19 pandemic and a post-pandemic era for humanity. Although the pandemic has already ended societally, but it still has a wide and long-term impact in the industry.
At the same time, however, the Chinese mainland is still under threat from the new coronavirus and influenza A virus, a reality that our team members cannot escape. We set out on this year's iGEM Journey, starting with the problem of our own partners, which is how to quickly and easily determine the type of infection to guide follow-up drug treatment.
During the covid-19 pandemic, there has been a significant increase in public health awareness. Our knowledge, understanding, and trust in the kit grew, so we eventually positioned our project as developing test kits.
On March 5, 2023, the World Health Organization (WHO) announced that the COVID-19 pandemic no longer constituted a "public health emergency of international concern." This declaration signifies a crucial step towards ending the global COVID-19 pandemic and officially entering the post-pandemic era.
However, simultaneously within mainland China, the threat of both the novel coronavirus and influenza type A virus persists, which our team members cannot overlook. Stemming from the challenges faced by our own partners, such as how to quickly and easily determine the type of infection for guiding subsequent medical treatment, we embarked on our iGEM journey this year.
Design process
Design Objective
Our team's goal was to design a universal modular precision disease diagnostic device with the following features:
- Accurate judgment: It is hoped that the equipment can carry out multi-objective comprehensive analysis to improve the accuracy of judgment.
- Easy to use: the tool can not only detect the target, but also automatically complete the target analysis, output easy to understand the judgment results, easy to use by the public.
- Noninvasive: Testing procedures should not include introduction of the instrument or engineered bacteria into the body.
Nucleic acid targets have been used in the diagnosis of viral infections, genetic diseases and tumors, showing good versatility and universality.
DNA molecular computing is a new type of computing based on the physical and chemical properties of DNA molecules.It has the characteristics of fast, parallel, miniaturization and low energy consumption, and has great prospects in the field of nucleic acid detection. Using DNA molecular computing techniques to detect nucleic acid targets is suitable for our goal.
Wet lab: Construction of the DNA molecular computing system
Selection of Molecular Computational Models - Neural Networks
In synthetic biology research, logical circuits often draw inspiration from the concept of logic gates in electronic circuits, achieving logical functions through the combination of various genetic components and regulatory mechanisms. However, understanding life through an analogy with electronic circuits overlooks the uniqueness of the signal systems in biology. Molecular signal systems in living organisms are entirely different from simple electronic circuits; their operational logic is highly complex, much like the intricate information processing that occurs in neural networks formed by the connections of axons and dendrites in neurons. Perhaps it is only by referencing life itself that we can truly comprehend life and develop more practical synthetic biology logic circuits.
Boolean logic is one of the fundamental concepts in the early stages of DNA molecular computing.Stratification of Boolean functions can solve complex classification problems in DNA molecular computing, but is not naturally adapted when dealing with inputs such as continuous biomarker concentrations.
In contrast, programmable neural networks with variable weights and thresholds can be (re) programmed for a given classification problem without changing the connections.
These properties enable programmable neural networks to perform complex linear and nonlinear data classification using compact networks, suitable for computational tasks with a large number of input biomarkers without loss of experimental realism.
Therefore, we decided to use a single-layer perceptron-based in vitro biochemical circuit for the detection and analysis of multiple nucleic acid targets through DNA molecular computing.
Constrained by the current state of DNA molecular computing technology, we can currently only construct a single-layer perceptron network structure using DNA probes as building blocks. However, enhancing the sensitivity and specificity of DNA probes will open up possibilities for creating more intricate multi-layer perceptron network structures, enabling more complex computations.
Dry lab: Target selection and dynamic simulation of the molecular computing process
Our first task is to build up the linear regression model, involves collecting and preprocessing mRNA microarray data, annotating and performing quality control on the samples. The dataset is splinted into training and test sets, and consider the severe class imbalance problem by using oversampling to balance the sample sizes across categories.
On this basis, various linear machine learning models are applied (such as linear regression, SVM, LASSO regression, etc.) on the training data, optimizing model parameters through cross-validation to select models with lower test error.
Next, the regression coefficients given by the model are converted into integer weight matrices, while strictly accounting for weighting, retaining only integerization schemes with minimal impact on the results.
Then, the above process is repeated until we obtain multiple models meeting the experimental requirements.
Finally, the results are outputted on Excel sheets, providing references for the design and validation of subsequent experiments.
Furthermore, we constructed the kinetic characteristic curve of the core component Bst DNA polymerase in the DNA molecular computing system based on chemical kinetics theory.
Feasibility analysis
The DNA molecular computing system consists of detection module, weighted summation module and fluorescence reporter module. The function of these modules is realized by strand displacement reaction between special DNA molecular probes.
In general, the system outputs the concentration signals of multiple nucleic acid targets as fluorescence intensity signals after weighted summation operation.
The fluoresce signals are measured to predict the concentration change in the reaction network, ODE45 is used to analysis the data. Fminsearch is used to compare the fitting data with the experimental data, and searched for the minimum value of the difference between the simulated curve and the experimental curve, so as to compare the fitting degree between the global simulated curve and the experimental curve, the curve that best approximates the true dynamics is obtained.
Each t-time corresponds to a fluorescence point, with experimental and fitting values. DISP (Kreal) can output the similar K value, make the difference minimum, can make the fluorescence curve, get the different substance concentration to the dynamics influence.
Apart from that, since the weights are expressed by the concentration of the released weight chain, it can be used as the theoretical basis for the study of the reaction mechanism to verify the positive correlation between the weight and the fluorescence value. During the prevalence of the COVID-19 pandemic, humanity's understanding of public health has greatly advanced. Our comprehension, awareness, and trust in diagnostic kits have also grown.
We work together as BUCT-China to make our ideas come true. As a result, we have ultimately designed our system, and developed a diagnostic test kit for our project.