From Biotechnology to Bioinformatics Software | Sebastian Ayala Ruano

The transcripts are edited for clarity, sometimes with AI. If you notice any incorrect information, let us know.

Sebastian’s Journey into Bioinformatics

Alexey: Hi everyone, welcome to our event. This event is brought to you by Data Talks Club, a community of people who love data. We have weekly events, and today is one of them. If you want to find out more, there is a link in the description. Click on it and check it out. (0.0)

Alexey: We have quite a few events in our pipeline. Do not forget to subscribe to our YouTube channel. This way you’ll get notifications about all future streams like the one we have today. Last but not least, we have an amazing Slack community where you can hang out with other data enthusiasts. The link is also in the description. (17.0)

Alexey: During today’s interview, you can ask any question you want. There is a pinned link in the live chat. Click on that link, ask your questions, and we will cover them during the interview. (34.0)

Alexey: That’s the usual introduction. I have the questions on the left side and Zoom on the right side, so I’m ready to start. Sebastian, are you ready? (53.0)

Sebastian: Yes. (1:08)

Alexey: Hi everyone. This week we are joined by Sebastian, a bioinformatics software engineer whose career bridges biotechnology and programming. Sebastian began in biotechnology engineering but realized research wasn’t his path. Through a minor in software engineering and a master’s degree combining both fields, he transitioned into bioinformatics software engineering. (1:09)

Alexey: I will ask you more about these things, and I’m really happy to have you here on this podcast. (1:36)

Sebastian: Thank you very much. (1:42)

Alexey: Let’s start with the first question. I always ask guests to tell us more about their career journey so far. I outlined it briefly, but I don’t know much about biotechnology or bioinformatics, except maybe one course I took on Coursera twelve years ago. So tell us about your career journey so far. (1:45)

Sebastian: I did my bachelor’s here in Ecuador. I studied biotechnology engineering, which is the application of living organisms in different areas. You can apply it to develop new potential drugs or in agriculture. All the applications are based on living organisms. Then, while I was doing my bachelor’s, I realized that I didn’t enjoy biology and wet lab work that much. (2:07)

Sebastian: I started to take some electives on programming and math, and I liked that approach more. I did a minor in software engineering, and in my master’s, I combined both fields in bioinformatics. During the last year, I have been working more on software development for bioinformatics. (2:47)

Alexey: You said you just finished your project in Denmark and moved to Ecuador to spend the winter there. Amazing idea. I live in Berlin, so I can imagine what it’s like now in Denmark. I think it’s even worse than in Berlin since it’s closer to the sea. Good idea spending time in Ecuador. (3:17)

Alexey: What exactly was your project about? Was it pure software engineering or bioinformatics? What do you actually do now? (3:41)

Sebastian: I did my master’s in the Netherlands in bioinformatics. My master’s thesis was at the Technical University of Denmark. During my thesis, I worked on a project called MCW2 Graph. It’s a knowledge graph to investigate the wastewater treatment microbiome. A microbiome is all the microorganisms that live in a specific environment. (3:53)

Sebastian: We collected samples from wastewater treatment plants and sequenced the DNA or proteins inside these environments. After processing this data, you can see the presence and abundance of microorganisms in these samples. We took studies from different regions and combined the data. Then we inferred microbial association networks based on abundance patterns. If two microorganisms co-occur in similar abundance, we infer they might be interacting. (4:20)

Sebastian: We created these networks and built a knowledge graph on top of them with additional metadata. That was my first project there, and then I worked on others. (5:17)

From Wet Lab to Computational Biology

Alexey: I’m taking notes so I can ask you later about these things. I know nothing about biotechnology. When COVID happened and companies started developing vaccines, was biotechnology involved there? How exactly does it work? (6:02)

Sebastian: Yes. In biotechnology, you do experiments in the lab to test different vaccines and how compounds can combat bacteria or viruses. That is biotechnology. You design components that are more efficient in combating viruses and bacteria. Bioinformatics helps reduce the number of experiments needed in the lab. (6:27)

Sebastian: You can run simulations or analyze the data to propose potential candidates that lab researchers can test. This reduces time and helps identify better options for experimentation. (7:08)

Alexey: I see. I think you have a slight problem with sound. Are you using your headphone microphone? (7:28)

Sebastian: Yes. (7:34)

Alexey: Can you try now? I cannot hear you. Sorry everyone, the sound is breaking a little. (7:43)

Sebastian: Now I switched to my laptop. (7:58)

Alexey: I think it’s better. The sound was breaking every few seconds, probably a Bluetooth issue. Let’s try now. You were saying that this is what biotechnology professionals do, especially when working with viruses and bacteria. (8:04)

Wet Lab vs Dry Lab Explained

Alexey: In the lab, they run experiments to see what helps combat these microorganisms. That’s what biotechnologists do, right? They work in companies like Pfizer or other pharmaceutical firms, running experiments. Why is it called a wet lab? Is it literally wet there? (8:23)

Sebastian: Yes, it’s to distinguish it from a dry lab, which is computational. In the wet lab, you physically do experiments and get your hands wet, so to speak. (9:00)

Alexey: So dry lab means you sit in front of a computer, and wet lab means you do experiments. When I imagine someone developing a vaccine, I picture them in white coats, wearing glasses, and holding test tubes. (9:24)

Sebastian: Test tubes, exactly. (9:44)

Alexey: Yes, doing these things with charts behind them. You didn’t enjoy that part, right, working in the lab? (9:47)

Sebastian: No, not really. I didn’t like it that much. The biological courses in my bachelor’s were too theoretical. They required a lot of memorization, which I didn’t enjoy. (9:57)

Sebastian: In the middle of my bachelor’s, I realized I didn’t like it much and decided to try something else. I took programming and math courses and realized I liked that approach better. In my first programming course, the first class was theoretical, but the rest were hands-on exercises. I liked that much more. (10:19)

Alexey: And in the Netherlands, where did you study? Which university? (10:46)

Sebastian: At Maastricht University. (10:53)

Alexey: I don’t know that one. (10:59)

Sebastian: It’s near the border with Belgium and Germany, next to Aachen. (11:01)

Alexey: Okay, I didn’t know that existed. So you studied there, and as part of your studies, you went to Denmark to work on projects? (11:06)

Sebastian: Yes. (11:19)

Alexey: Okay, and we talked about biotechnology engineering. These are the people who develop drugs and vaccines, right? So that’s biotechnology. What is bioinformatics? How is it different? (11:21)

Sebastian: Bioinformatics is data science applied to biology. It’s using data science techniques to analyze biological data. We take information generated in the lab and perform typical data science processes like exploration, analysis, and modeling to interpret results and make predictions. (11:45)

Alexey: So basically, you do data science too, but biotechnology is an older field, while data science is about fifteen years old. (12:24)

Sebastian: Yes, that’s true. (12:30)

Bioinformatics as Data Science for Biology

Alexey: Okay. What kind of data do you have? I mentioned that I took a course on bioinformatics years ago. I think we used R, and the data consisted of four different letters. What are these letters? (12:35)

Sebastian: These are nucleotides. Genomic data, like DNA, is made up of chemical compounds called nucleotides: adenine, guanine, cytosine, and thymine. These four letters form the DNA sequence that carries genetic information to make proteins responsible for biological functions. (12:59)

Alexey: I remember the data looked like long strings of text with those four letters. You mentioned sequencing DNA and proteins. What does sequencing mean? Aren’t they already in a sequence? (13:49)

Sebastian: Sequencing means reading and decoding the DNA from a sample. The sample could be environmental or biological, like blood or saliva. (14:16)

Alexey: A few years ago, I used a service called MyHeritage to learn where my ancestors came from. They sent a small swab kit to collect saliva, which I sent back. That was the sample, right? (14:23)

Sebastian: Yes. (15:05)

Alexey: After a few weeks, they sent results showing I’m mostly from the Baltic countries, partly Western European, and a mix of others. That’s one of the applications of bioinformatics, right? (15:10)

How DNA Sequencing Works

Sebastian: Yes. They take your sample and sequence it using machines called sequencers. The DNA is extracted, broken into small pieces, and decoded to determine the order of nucleotides. (15:30)

Sebastian: Then they assemble everything to reconstruct your whole genome. From that, they compare it to databases linking specific sequences to genes, proteins, or populations. Biomarkers in your DNA indicate ancestry or health traits. (16:25)

Alexey: I think one part of my DNA showed I was one percent Irish or Welsh. They must have matched a DNA segment to Irish populations in their database. (17:16)

Sebastian: Exactly. They compare your data to a reference genome. (17:30)

Alexey: Interesting. I assume you didn’t work on ancestry projects but on something else. Now that I understand more about biotechnology and bioinformatics, could you explain again what you did with the microbiome project? (17:56)

Sebastian: Yes. That was my master’s thesis project. We worked with metagenomics, which studies DNA from environmental samples instead of individual organisms. (18:16)

Sebastian: For example, we could take a sample from a lake or soil to see what microorganisms live there. We decode all the DNA and reconstruct the genomes of all microorganisms in the sample. This is more complex than a single blood sample since it contains many organisms. (18:43)

MCW2 Graph and Wastewater Microbiomes

Alexey: So the goal is to understand what kind of organisms are in the sample, right? (19:29)

Sebastian: Yes. In our case, we focused on wastewater treatment plants. We analyzed data from different locations because there were many available samples. (19:41)

Sebastian: We downloaded abundance tables, where rows represent microorganisms and columns represent samples, with counts showing how often each organism appears. We combined datasets from multiple studies and categorized them by biome. Then we inferred microbial association networks using correlations between microorganisms’ abundance patterns. If they often appear together, we assume some form of interaction. (20:10)

Alexey: And what does co-abundance patterns mean? So this when you have this sample, right, and you understand that you have like I don't know three samples of three counts from organism A, five counts from organism B, and then there is in some other sample you have kind of similar thing, right? And then by analyzing all this abundance tables as you call them, right, you can see that these organisms tend to coexist, right? So you see them together often in the same environments, right? (21:12)

Sebastian: Mhm. Yeah. Okay. Okay. Yeah. That's basically it. And yeah, we kind of create these links between these microorganisms if they are co-abundant. (21:52)

Alexey: Interesting. For me, I never worked with this kind of data, but what I did is a bit of NLP. In NLP, you have this co-occurrence tables. When you have some context and you see that certain words appear together, then you can say that they are co-occurring. (22:05)

Alexey: There is a special term for that I don't remember, like New York, right? New York is always like that. If you see New York, it's very likely that they co-occur. Or names of movies, or some expressions, right? If you see one word, another one is likely to happen nearby. (22:31)

Building Microbial Networks with CC Lasso

Alexey: Sometimes certain words just follow each other. Like after a verb, make, we have another noun like make noise, right? Maybe this is similar to what you do. This is what they do in NLP, and then they use simple statistics to calculate co-occurrence. (23:10)

Alexey: I remember it involved logarithms, but basically you do a statistical test to check how likely it is that these two things just happen to appear together. You find if it is statistically significant that they co-occur. Then you perform the test and can conclude that this is a bi-gram or pattern. (23:31)

Alexey: In recommender systems, you can find co-occurrence patterns too. For example, if I buy a phone, I might need a charger. Certain items are often bought together, and we can find these patterns in transactions. It’s a similar statistical approach. (23:50)

Alexey: And what you do looks very similar, right? What do you use there? (24:23)

Sebastian: In that case, we use a method based on something called CC Lasso. It is based on the Lasso algorithm to predict potential interactions among the microorganisms. In the end, we calculate correlation values among every pair of microorganisms. (24:31)

Alexey: So how does it work? Do you create like a regression model? With these interactions between organism A and organism B, right? If the coefficient is not zero, then it means that these things are co-abundant, right? (24:58)

Sebastian: It's like we obtain a value between minus one and one, and then we set a threshold to define if the value remains in the table or not. Then we keep the significant values, and we have negative and positive correlations. These can have biological interpretations, but we should be careful with them. (25:22)

Alexey: What does the negative correlation mean in this case? (25:55)

Sebastian: It means that when one microorganism is in the sample, the other is not there. Maybe they release some components that prevent the other from living in the same environment. (26:01)

Alexey: I see. So positive correlation means they like to coexist together, and negative correlation means that they don’t exist together, right? (26:21)

Sebastian: Exactly. Yeah. (26:28)

Alexey: But could it be just geographic? When you do this co-abundance pattern extraction, do you restrict yourself to a certain area? Let’s say you take a lake, take all the samples around this lake, and then do the analysis there. Maybe things potentially coexist, but in your samples you don’t see them. (26:30)

Sebastian: Yeah, it can be. (26:48)

Protein–Ligand Simulation Basics

Alexey: That’s interesting. So did you have experience working with non-bioinformatics data before? (26:54)

Sebastian: In general? (27:00)

Alexey: In general, yeah. (27:03)

Sebastian: Actually most of the projects I’ve been working on have been in biology. Different types of biological data. At some point I worked more on simulations of proteins. (27:06)

Sebastian: In that case, you have proteins and ligands that interact, and then you run simulations. You simulate a water box, let the proteins move, apply physicochemical parameters, and then measure interactions. I did that during my bachelor's. Most of my work is biology related. (27:26)

Alexey: The reason I’m asking is because I was curious if you observed any similar patterns. Right now you were talking about co-abundance patterns, and I thought this is very similar to what I was doing before. (28:00)

Alexey: It made me realize that my past experience working on something unrelated could actually transfer to bioinformatics. For me, that was really cool. So I was wondering if there are other parallels like that. (28:17)

Alexey: That’s why I asked about your experience working on other types of data. I imagine someone listening might have similar thoughts. I see that there’s Mark who says he’s a microbiology student and an engineer. (28:36)

Alexey: Hi Mark. Anyway, I know we prepared questions, but I’m sidelining a bit because I’m really interested in these things. You mentioned protein experiments and simulations of proteins. (28:53)

Alexey: Was it a few years ago when Google managed to use AI for protein structure prediction? (29:15)

Sebastian: Oh yeah, AlphaFold. It predicts the structure of proteins from their sequences. (29:28)

Alexey: For me, everyone was talking about this as a breakthrough. Nobody could do it before. Since my background in this is limited, I didn’t understand the significance. (29:34)

Predicting Protein Folding in 3D

Alexey: Now I get a chance to talk to someone who understands these things. Can you explain in simple terms what’s the big deal about AlphaFold and why it matters? (29:58)

Sebastian: Basically, proteins are really important because they perform all the functions that living organisms do. (30:05)

Alexey: So for me, I have this mentality of a gym bro. Proteins are important because after training you need to consume proteins. (30:17)

Alexey: But they are important on other levels too, right? Not just for that. (30:32)

Sebastian: Exactly. They are small molecules that perform all the jobs at small and large scales. There were many efforts to understand them because predicting their structure from their sequence is complex. (30:37)

Sebastian: The sequence is a string of amino acids, and when proteins perform their functions, they fold into 3D structures. This folding determines their function. (30:58)

Alexey: Okay, so you have a string, and when you put it into real space in 3D, it starts folding. The same sequence can fold differently and serve different functions, right? (31:27)

Sebastian: Exactly. The search space is huge, larger than the number of stars in the universe. Predicting the 3D structure from the sequence is extremely difficult. (31:46)

Sebastian: There were many traditional approaches, and there’s a competition called CASP that evaluates these prediction methods. When AlphaFold entered, it performed far better than other teams. (32:15)

Sebastian: Knowing the structure from the sequence is crucial because doing it experimentally in a wet lab is very time-consuming. My friend’s PhD project was dedicated to determining the structure of one protein. (32:50)

AlphaFold Revolution in Protein Prediction

Sebastian: It takes years, but AlphaFold can do it in seconds. It’s incredible. (33:30)

Alexey: Instead of five years. (33:42)

Sebastian: Exactly. Depending on the type of protein, simpler ones can be predicted with about 90 percent confidence. More complex ones still need experimental validation. (33:48)

Alexey: What are the practical implications? Can we do drug discovery faster now? (34:21)

Sebastian: Yes. We can do drug discovery and explore protein applications in different fields like agriculture and industry. Enzymes can accelerate specific processes. Applications are everywhere. (34:27)

Alexey: And these proteins are important because they’re building blocks for many organisms, right? (34:56)

Sebastian: Yes. (35:02)

Alexey: That’s why after training you consume proteins they’re the building blocks. The body deconstructs them and uses their parts for its purposes, right? (35:02)

Sebastian: Yes. There are different types of proteins. Some assemble into complex structures, while others perform functions like carrying oxygen, such as hemoglobin. Every biological function involves proteins. (35:27)

Alexey: And living organisms consume these proteins, and the body uses their parts for its needs, right? (36:00)

Alexey: Interesting. I can’t say I understood everything, but it’s clearer now. Thank you. (36:13)

Alexey: You worked on quite a few projects. The document mentions MCW2 Graph, VueGen, and VueCore. What are these projects and what do they do? (36:20)

Sebastian: Yes, I already mentioned MCW2 Graph. It’s the knowledge graph with microorganisms. (36:40)

Inside the MCW2 Knowledge Graph

Alexey: Last year we talked about co-abundance patterns. Once you extract the patterns, you can put them in a graph. These nodes have positive coefficients, these have negative ones, right? (36:45)

Sebastian: Right. Every node is a microorganism, and there is an edge if there’s a co-abundance pattern between them. We created these graphs and enriched them with metadata. (37:03)

Sebastian: For example, we know a microorganism releases a specific metabolite or lives in a specific biome. We added that information to build a knowledge graph. (37:20)

Sebastian: Then we applied clustering algorithms to find microbial communities involved in specific pathways or biological processes, important in wastewater treatment and other fields. (37:33)

Alexey: I just Googled MCW2 Graph and saw your article “Building a Knowledge Graph of the Wastewater Treatment Microbiome.” You’re the author, that’s great. It describes it as an open-source knowledge graph. (37:55)

Alexey: That’s interesting. How can people use it now? (38:24)

Sebastian: We created a Streamlit web application with exploratory data analysis and microbial association networks for every biome. People can download the dump file and open it in Neo4j. (38:31)

Sebastian: They can also access the raw CSV file or visualize it on the web application. (38:53)

Alexey: When I have this Neo4j database, what can I do with it? (39:07)

Sebastian: You can run graph algorithms like clustering or centrality analysis. You have the experimental edges we inferred and the extra metadata. (39:13)

Alexey: So you can analyze things you didn’t directly observe but that the graph structure suggests. You can infer potential co-abundance relationships, right? (39:37)

Sebastian: Mhm, exactly. (39:54)

VueGen: Automating Scientific Reports

Alexey: That’s really interesting. And the other two projects, VueGen and VueCore? (39:54)

Sebastian: That’s what I worked on last year, after my master’s. VueGen is a Python package that automates the creation of scientific reports. (40:00)

Sebastian: You provide a directory with a specific structure containing tables, plots, and files. It can handle static or interactive plots, network data, or HTML files. With one command, you can generate documents like PDF, HTML, DOCX, presentations, or Streamlit web apps. (40:14)

Sebastian: The idea is to make it easier to create these reports. (40:41)

Alexey: Yeah, I just shared the links to MCW2 Graph and VueGen in the chat. I also put them in the description. (40:48)

Alexey: It only has 28 stars. Let’s increase that number. Everyone watching, please do the same. (41:00)

Alexey: Okay, now you have a star. What do you use inside? Do you just parse the data, or do you use something intelligent like an LLM? What’s happening inside? (41:11)

Sebastian: Right now under the hood we are using Quarto for the Quarto and Streamlit projects. This is traditional object-oriented programming. We have not yet added large language models to interpret the information, but that is a project we plan for the future. Basically, we interpret the directory structure and then create a YAML file with the information, including paths, descriptions, and other details. For Streamlit apps, we generate Python scripts for each tab of the web application, and for Quarto we create a QMD file, which is a markdown file that Quarto can render into PDFs or other formats. (41:25)

Alexey: I see it links dependencies. VueGen uses Quarto to generate various report types, and when I click, I can learn more. I had no idea this library existed. I also noticed that VueGen is available on Bioconda. What is Bioconda? (42:29)

Sebastian: Bioconda is basically a branch of Conda that contains packages related to biology, bioinformatics, bioengineering, and similar fields. (42:48)

Alexey: Are there many libraries in Bioconda? I see there are 11,733. That is a lot. People have been busy creating biological packages for Anaconda. I had no knowledge about this whatsoever. It is interesting. So I am thinking of a direct application someone working in a lab doing experiments. With this tool, they can point to a folder and generate a report. (43:02)

VueCore: Visualizing OMIX Data

Alexey: The report can be in whatever format, such as Streamlit, Jupyter Notebook, or PDF. That is nice. The other view you mentioned was it Vuecore? (43:56)

Sebastian: Yes, Vuecore. The idea is to make it easier to generate interactive and static visualizations from omics data. Omics data refers to things like genomics, proteomics, and metabolomics. Proteomics focuses on proteins, genomics on genes, and metabolomics on metabolites. You take samples and identify the peptides and proteins in them. Then you can analyze the different layers of data and integrate them to have a holistic view of a living system. (44:09)

Alexey: Interesting. I see that we have a few questions. Maybe we can cover them before we run out of time. Yekaterina is asking which beginner-friendly bioinformatics projects would best showcase her skills to future employers. If someone wants to work in bioinformatics, what kind of projects should they do for their portfolio? (45:08)

Sebastian: I think what helped me during my bachelor’s degree was finding a topic that I liked and playing with packages from Bioconda. There are many tools for parsing data and strings. It is easy to test these packages and apply them in a specific domain that interests you. You could also create a simple Streamlit web app to demonstrate what the tool does. When I was in my bachelor’s program, many biologists were doing things manually. Simple tasks like creating a batch script to parse files made a huge difference. (45:42)

Sebastian: Things that took an hour in a graphical interface could be done in a minute on the command line. I really liked discovering that. Playing with the command line and commands like grep or cut helped me a lot. For a bigger project, you can explore Bioconda or Bioconductor in R. The bioinformatics community is still larger in R, and there are many datasets available. You can test the tools, run exploratory analyses, and showcase that work. (46:53)

Using AI and LLMs in Bioinformatics

Alexey: Do you use AI or large language models in your work these days? For me as a developer, AI helps tremendously, so I assume it is similar in bioinformatics too, right? (47:50)

Sebastian: Yes, large language models and generative AI are becoming very important in this field. Many people are working on tools to integrate and improve them. I personally use them a lot for brainstorming ideas or writing documentation. In my recent projects, we also wanted to implement MLOps protocols to test how they would perform. (48:11)

Alexey: Amazing. Another question: which online resources, tutorials, or courses would you recommend for biochemistry graduates interested in biotech and bioinformatics? Biochemistry sounds like an intersection between biology and chemistry, right? (48:54)

Sebastian: Yes. I would say right now I do not remember a specific resource, but I know there are good curated lists on GitHub. There is one called Awesome Bioinformatics that includes many useful courses and tutorials. (49:14)

Alexey: Amazing Bioinformatics? (49:53)

Sebastian: Yes, Awesome Bioinformatics. (49:55)

Alexey: I see this curated list of awesome bioinformatics resources. (49:57)

Sebastian: Exactly. (50:02)

Alexey: It has 3.7 stars. I will share it, but it is easy to find by searching online. Multiple people have starred your repository. Thanks very much. (50:08)

Sebastian: Thank you very much. (50:18)

R vs Python in Bioinformatics Tools

Alexey: Mark says he uses Bioconda too. I did not know about this. So Bioconductor is in R, and Bioconda is similar but for Python, right? (50:25)

Sebastian: Yes. The community is still bigger in R, but more people are transitioning to Python. Most packages are still published in R. Sometimes bioinformatics tools are developed by scientists who are not trained software developers, so when you try to scale those tools, you might find issues. (50:40)

Alexey: And you need to scale because you work with large amounts of data, right? When those libraries are not scalable, what do you do? Do you rewrite them in Python? (51:12)

Sebastian: It can be complicated. Sometimes you need to expand the source code or contact the developers to ask for help with a specific use case. It can be tricky. (51:35)

Alexey: I wonder if large language models can help with that. With coding agents, you can easily translate from one language to another and even fix bugs. For example, I can take a project with no tests and ask a code assistant to write tests for it. It tries to understand the project and generate them automatically. That is really cool. What are you working on now? (51:53)

Sebastian: My last project was Viewer, which I just finished. The idea was to support different plotting libraries for visualizations. (52:26)

Alexey: I found a website called VueCore. (52:50)

Sebastian: No, it is Viewer. (52:58)

Alexey: How do you write it? (53:00)

Sebastian: V-u-e. That one is more focused on you. (53:04)

Closing Thoughts from Ecuador

Alexey: Ah, the one we talked about earlier. Sorry, I misunderstood. I did not hear it clearly before. You are now relaxing in Ecuador, right? (53:17)

Sebastian: Yes. That was my last project, and now I am taking a break here in Ecuador to enjoy the mountains. I really like hiking and climbing. In Denmark, I did not have that opportunity, so here I am making the most of it. (53:36)

Alexey: Are there many parrots? (53:51)

Sebastian: Many what? (53:58)

Alexey: Parrots — the birds. (54:00)

Sebastian: The birds? Yes, a lot. Birdwatching here is amazing, especially in the Amazon jungle. (54:03)

Alexey: That must be amazing. I actually have another event starting soon, so I need to go. Sebastian, thanks a lot. It was really nice talking to you. I learned many new things. I suspected proteins were important not just for the gym but for other purposes too. Thanks for explaining that. I also finally understand the buzz about this on social media. (54:10)

Alexey: Thanks for joining us and sharing your insights. For me, the aha moment was realizing that with my experience, I can actually relate to what you did. That was interesting. Thanks to everyone for joining and asking questions. Give a star to Sebastian’s project and also to Viewer since it is open source. (54:49)

Sebastian: Yes, it is open source. Please give a star to that too. Thank you very much, Alexey. I had a lot of fun. (55:13)