This being our project’s first year, our team has focused on setting a solid foundation from which future teams will be able to continue with it. This purposefully is an initiative that we have planned to finalise in subsequent iGEM editions, as it necessarily needs to be concluded in a larger time span.
Our team has started this journey by ensuring the validity of our project and setting some key starting points for future wet lab steps. The validity of our project has been hypothesised by our team members, and the market niche for our potential product has been analysed by our data scientist by market researching. Both of these have been proven via survey launch and analysis of the data retrieved from them.
As our project was focused on a dietary supplement to treat anxiety and depression we wanted to see the current situation of these mental illnesses. Thus, a survey was made to examine Spain’s mental health. Not only were we keen on knowing its current situation, but also, its evolution before and after the pandemic, and, mainly, the lockdown.
We defined three main hipotesis to corroborate our suspicions:
Hypothesis 1: Mental Health has deteriorated since the COVID-19 pandemic.
Hypothesis 2: Society is nowadays more aware of mental health than 5 years ago.
Hypothesis 3: Society’s habits and dynamics have changed due to the COVID-19 pandemic
Regarding the first hypothesis, we found in questions five and six, related to the first hypothesis, how the percentages of agreement and disagreement reflect the worsening of the respondents' mental health.
Also, we studied with a factorial analysis both hypothesis one and three as a whole. This analysis suggested that questions 13 and 14 had great importance in an individual’s mental health based on the factors and characteristics of the model. Notice that the importance of the question is determined by the length of the arrow in the plot below.
Factors from one to six, related to nutrition, sleep, sports habits and so on, had a differential weight when compared with habits related to addiction and narcotics. We suspect an error related to social convenience here, as two last topics are a touchy subject for most people.
The following figure shows how responses related to good habits decrease in some of the factors discussed (in green), thus increasing the number of responses related to bad habits (in red and orange) with the passage of the pandemic.
However, contrary to what we suspected, as the figure below suggests, respondents are not used to talking about their mental health with whom, at first, would seem to be their closest people. This response, however, only reinforces our initial thought when we launched our project: Mental health is still a taboo topic and it needs to be acknowledged as any other disease for it to be treated with more ease.
We defined a second survey to delve deeper into aspects of interest that we did not address in the first one, such as:
Hypothesis 1: Whether the tools provided and current education around mental health issues are accurate.
Hypothesis 2: Whether a respondent would be willing to take a supplement or drug to treat a clinical depression or similar condition.
Hypothesis 3: Whether a respondent would be willing to take a supplement or product from the circular economy.
Further data retrieved from our second survey however showed that this would be this way due to the lack of information and resources people have to treat these disorders (figure below).
Although they do not feel they have adequate resources, most respondents are willing to treat their disease with therapeutics.
Similarly, the majority of respondents believe that these treatments are beneficial to their health, with only 18% of respondents being indifferent.
Regarding circular economy initiatives, we have deduced that, in general, people are in fact positive about circular economy and its potential applications (see figure below). So we are one step further to confirm that our product has, indeed, a place. Further questions in our survey dove deeper into how familiar society is to circular economy strategies and the vision they have on them.
In fact, our data suggests that there is a believe in our society about circular economy solutions being beneficial (see figure below). This result was partially expected, as these have been widely advertised and have been of great impact lately. However, we did want to be sure about the potential impact our product would have once launched, when it comes to its awareness and social rejection.
Moreover, we have seen a general willingness to try solutions that have emerged from circular economy solutions (figure below).
Nevertheless, we do see a decrease when it comes to the willingness to take a medicine or supplement derived from circular economy strategies (Figure below). However, we lack data indicating whether this behavior originates from the general apprehension against medicines or from a distrust of circular economy solutions in these areas in particular. But, noticing the tendencies from previous questions, we could infer that it is more probable to be as it is due to a general uneasiness to medicines (data not shown).
After analyzing our first survey we were keen on extracting more abstract knowledge. This is why we developed a model to further learn characteristics of our data. Due to the nature of the survey, the most appropriate category of learning was clustering. To perform this technique, we needed to study the correlation of the data, and process the data, for the training phase of the algorithm. We use the topological algorithm DBSCAN to find groups in the data after applying principal component analysis.
First, we select the principal components on which we are going to apply the clustering algorithm. It can be seen that when projecting the data on the first component and any of the other two we do not have a clear division of the data.
Although, when projecting on the second and third component there seems to be a division into two groups
Having selected these two components, we need to see what relationship there is between them and the original variables, that is, the survey questions.
First, the third principal component does not have significantly high correlations with the questions, which may indicate indecisiveness or that responses are being given for convenience.
First, the third principal component does not have significantly high correlations with the questions, which may indicate indecisiveness or that responses are being given for convenience. While the second component has correlations with questions 13 and 14, which refer to the habits of the respondents. We see the same pattern in these questions, indicating that habits did not change before and after the pandemic. This implies that the component represents individuals with good habits.
Because the projected data are anisotropically distributed, we used the DBSCAN algorithm which is robust to this problem. Various values for the epsilon parameter, which controls the radius of neighbors to look at, were tested and corroborated through the silhouette coefficient and the silhouettes of the clusters.
The highest value for this coefficient was 0.4255, which corresponds to an epsilon of 0.18 and a classification into two clusters, as we suspected after looking at the scatterplot of the data.
Finally, the clusters have been characterized as people who are slightly uncomfortable discussing mental health issues with family and friends (0) and people who are very uncomfortable discussing mental health issues, mainly in the work environment.
Analyzing the mean of this variable we see that in cluster 0 there is a tendency to have less discomfort in talking about mental health problems as opposed to cluster 1. In addition, in job/classmate in cluster 0 there is a tendency not to have discomfort with work/classmates while in cluster 1 there is more reticence, this is related to what was commented with the previous variable. In the rest of the variables the differences are not very significant, so they do not represent an element to differentiate the individuals. Individuals marked as -1 do not belong to any cluster.
The main objective this year was to provide a firm starting point for the next steps in the project to be taken with more ease. Thus, our team has designed functional base DNA constructions with the two main enzymes that participate in serotonin biosynthesis in humans: Tryptophan hydroxylase (TPH) and tryptophan decarboxylase (TDC).
The cloning strategy to introduce TDC into the pET28a vector was based on conventional restriction enzymes (IIR). In this case, EcoRI and HindIII were used, one at each end. This double digestion ensures that the insert was placed in the correct direction and not in reverse of the reading pattern.
The insert consists of the human TDC sequence optimised for Escherichia coli, flanked by the corresponding restriction sequences, in 5' to EcoRI and in 3' to HindIII, previously added by PCR. In this way, by means of a previous digestion of both the insert and the vector, complementary ends are obtained in both, and by means of the T4 ligase, they are joined.
The cloning strategy performed to obtain the TPH protein of interest is based on the infusion method, whereby by creating complementary regions between the ends of the insert and the pET15-MHL, and treating the mixture with the corresponding enzymes, a directional recombination is achieved. Insertion of the TPH sequence involves replacement of the SacB gene, which provides negative selection of the original plasmid if the treated bacteria are grown in 5% sucrose-enriched media. Prior to the insertion process, the vector is linearized by the enzyme BseRI whose restriction targets are located at positions 5381 and 7370. The employed sequences are available in Table 1.
Cloning of the pET28a-TDC construction was done by means of standard molecular biology techniques (see Engineering page in our wiki). Briefly, pET28a vector was extracted from the strain containing it. Vector and insert were processed to perform the ligation reaction, and the final construction was electroporated into E. coli TOP cells for long term storage.
Construction of the pET15-MHL-TPH construct did not result possible. This was mainly, as we suppose, due to our team not being familiar to the technique. In synthetic biology issues, this could mean a poor understanding when it comes to the DNA fragment design, problems regarding experimental conditions and repetitivity, or just the need of more time to assimilate its steps. Sadly, our team lacked the time to diagnose what was going wrong with their technique, so in the end we had to abort the experiment.
Table 1: Sequences used in wet lab
| TPH optimised for its production in E. coli | URL or sequence (5’ to 3’) |
|---|---|
| TPH optimised for its production in E. coli | Link |
| TDC optimised for its production in E. coli | Link |
| Primer FwTDC | ATCCGAATTCATGGACATTGAG |
| Primer RvTDC | CCGCAAGCTTTTATTGAACGTC |
| Primer FwTPH | TTGTATTTCCAGGGCATGTGTGAAA |
| Primer RvTPH | CAAGCTTCGTCATCACTAGATCGACTGGTC |
Note that the constructions here obtained do not meet the Biobrick Assembly Standards. Our lab personnel were unfamiliar with these strategies and, as a starting point, it was established that it would be better to store the constructions in common commercial vectors, more familiar to the ones used in the everyday procedures in our lab. The parts submitted to the iGEM Part Registry, however, have been done so that there are no modifications with regards to the coding sequences we have employed, but require the Assembly Standards characteristics. In fact, the main objective for future editions will be to compare different constructions that do meet the Standards to optimise enzyme production in relation to the more common vectors employed here (see Proposed Implementation page in our wiki).