Bioassay Data Analysis Platform for Researcher Efficiency
Advisor: Dr Zeynep Araci
Internship Location: Cataya Biotechnology
Onsite Mentor: Yang Zhang
How do we convert raw data to a graph in the lab? There are two methods: tediously apply a bunch of formulas to your data sheet, or use a data analysis tool. Currently, data analysis tools such as MyAssays are not very customizable and require a very rigid data input and only have data analysis automation for their select assays. By using Python through Jupyter and Flask, my project aims to create a platform which is fully customizable for biological data analysis, allowing all kinds of different assays and parameters to be analyzed. I will be working with Cataya Biotechnology, first researching on my own on how assays are analyzed and learning Python skills. After this period, I will begin writing the program and testing it on various data that has had a graph generated before and comparing the results between my program and the correct graph. If I have time, I will be looking into creating a mobile version and also incorporate AI and number recognition. This project gives me the great opportunity to combine my skills from the biology lab and programming, creating a tool that can convenience many researchers in their discoveries.
Project Blog
Week 0: Introduction and Background
Hello everyone! Welcome to my senior project. In this first blog post, I’ll be discussing the background of bioassay analysis, my project purpose, and my prior research.
Biological assays are widely used in biomedical and clinical research. An experiment can be set up (e.g. 96 well plate), which can be analyzed with various methods. Data returned can come in the form of absorbance values, fluorescence, etc. However, this data needs to be converted to a graph in order for it to be useful to the researcher. This can be done through tedious formulas and analysis through Excel/Sheets, or a software can be used to automate this process.
The programmed software aims to allow for simple and easy conversion of data values in various biologic assays into analyzable data (graphs, charts, etc.) instead of tedious conversion and graph generation via spreadsheets. I hope to create a software that can serve as easy graph generation and data analysis for biological researchers (and have a collection of data types to choose from), and allow it to be exported. I plan to write Python Jupyter notebook scripts to automatically process biological assay data including MTT, Cell Colony Formation, ELISA, etc. and visualize it on the go. Future versions will include deploying a Python app using the Python Flask framework so everyone around the world can use it. For assay protocols and program design, I will use publications from PubMed and Google Scholar to ensure the graph generation is accurate according to the nature of the experiment.
As a high school researcher in cancer biology and immunology, I am familiar with the workings and concepts behind many biological assays such as MTT, Colony Formation, ELISA, etc. I have experience conducting these assays and know how to extract data from them manually. I have also taken classes in neuroscience, biology, and chemistry at a college level, in addition to reading textbooks and papers about immunology such as Janeway’s Immunology and Campbell’s Biology. I have some experience in programming and computer science, and it will be a great opportunity to learn more about app design and data analysis in the biological context.
In the first few weeks, I’ll be researching selected assay implementations in code and learning the libraries and modules I’ll be using in this project. Stay tuned!
Week 1: Assay Studies and Implementation
Hello everyone! Welcome to Week 1 of my Senior Project. This week, I learned some basics with Jupyter Notebook, studied some assay implementation, and created a GitHub repository.
To start, I learned the basics of setting up Jupyter Notebook, and some basic packages such as the Pandas package. Pandas is a great data analysis package; one thing that it can do is read an Excel file and allow the data within to be manipulated and analyzed. I brushed up on some basic Python skills such as iterators, conditionals, lists, etc.
I also studied the theory of some assays that can be used for my project, including cell viability, colony formation, and ELISA, and studied the mathematical formulas used to analyze the readouts.
To get a head start, I successfully manipulated an Excel file for a basic cell viability assay. The most basic cell viability assay (MTT) is a colorimetric assay that measures absorbance values from the dissolution of colored formazan salts. After gathering the absorbance values, a researcher has to perform several calculations on the data to generate graphs. I was able to successfully automate the calculations, but I am yet to implement the graphs.
Next week, I'll be researching more assays and their analysis process, and collecting sample data!
Link to the project here.
Week 2: A New Direction
Welcome back to another SP blog. This week, I explored some assay implementation, such as the ELISA assay and clonogenic assays, and studied some statistical curves such as 4PL. However, I found some problems with my approach: there are thousands of different assays, and researchers are always creating more. It is not feasible to make a "preset assay" tool for my platform and I will instead focus on the "custom assay" tool.
While I was researching, I discovered a tool called GraphPad Prism 3D. This is a paid software in which users can input their data into a spreadsheet, and a variety of statistical analysis tools can be performed on the data, leading to beautiful graphs. This is similar to what I wanted to do for my project. I downloaded the free trial of GraphPad and started taking note of its features and tools. What I'm going to do is to first understand/evaluate and implement similar data analysis and graphing tools in Python (inspired by GraphPad) as a starting point for my own project.
Through this, I will be learning a lot of statistics and data science, so I'll be working more closely with my external advisor.
That's it for now... see you in Week 3.
Week 3: Basic Data Manipulation
Hello friends!
Let me tell you on what happened this week--
Taking inspiration from graphing tools on the market, I wrote Python scripts for several basic data manipulation functions. Of these are:
Transform and Transform Concentrations
Transform and Transform Concentrations allows for simple transformation of data. The X or Y values have simple functions applied to them, i.e. multiplication by a constant, addition, log, subtraction, etc. and the changing of units. Of these also include biochemistry functions, such as those used in Michaelis-Menten enzyme kinetics.
Normalize
Normalize fits data onto a common scale. Usually this can be done with percentages and fractions, with 0% and 100% defined by different means. Normalization is very important for a variety of bioassays, such as the MTT we saw two weeks ago.
Fraction of Total
Fraction of Total is a very simple function that divides values with a selection of totals (column total, row total, etc.). Confidence intervals can also be calculated.
Transpose
Transpose can switch rows to columns, and columns to rows.
They'll all be on the GitHub, and I'll see you next time.
Week 4: Regression
Hello friends!
Want to know what I did this week?
Not much regarding to my project, since we had a day off in the beginning. However, I have implemented some key functionalities in statistical analysis:
Linear regression, which can estimate a linear relationship and the values of variables based on other values, and logistic regression, an S-shaped curve which can estimate dependent variables between 0 and 1. I'll be implementing some non-linear regression curves on Friday, and besides the bug fixes, that's all I have! I will explain more in detail next week about my findings and progress.
See you next week...
Week 5: Halfway
Hello friends!
This week marks the halfway point. Aside from a day polishing and bug-fixing in preparation for an assessment by my wonderful SP Coordinator Ms. Nagpal, I had several objectives completed this week:
Nonlinear Regression: Similar to linear and logistic regression like mentioned last week, nonlinear regression aims to have a curve fit but the user must select the curve type. For example: polynomial, exponential, etc. This function is used in a lot of biological assays.
Area Under Curve: If everyone remembers calculus, the area under the curve can be represented with a definite integral. Commonly in science it can represent the cumulative effect of some experiment, like a drug, for example. This is useful if you want to measure accumulation.
After Spring Break (I won't be posting a blog during Spring break), I'll finish up the remaining functions and we can start implementing Flask and web development.
Thank you for reading!
Week 6: Hitting the Books
Hello friends!
This week was not super interesting in terms of execution, but there were some very interesting things learned in preparation for next week.
Handbook of Biological Statistics:
My advisor shared with me a very nice handbook for explaining concepts in biological statistics, including many statistical tests and methods. It was initially very difficult learning statistics, and this book helped a lot. Check it out here: https://www.biostathandbook.com/index.html
Learning the statistics here will be very important in implementing them in code. We'll be using SciPy's statistical functions.
Flask:
Flask is a Python web development framework which is great for small to medium projects such as this one. Since I'm new to this framework, I have been reading and looking at tutorials to get ready for the real work, which begins next week.
Thanks for reading!
Week 7: Thanks, Mr. Zhang!
Hello Friends,
This week was very exciting, since we're starting to make the final product! Through the Python Flask framework, we can build web applications with ease. In the beginning of the week, I followed along with Miguel Grinberg's Flask tutorial and practiced programming a microblog (similar to our senior project blogs!) If you are interested, check out the tutorial here: https://blog.miguelgrinberg.com/post/the-flask-mega-tutorial-part-i-hello-world
Then came in some difficulty. I'm kind of new at using Flask and web development, especially HTML. It was rather slow learning from online tutorials, until my external advisor helped me out with learning the programming. Turns out it is more simple when you break down the code into little parts. I'll be writing more functional code next week, and maybe soon we can also try making the platform visually appealing. The Beautiful Soup package can help with this, and maybe I can try experimenting with Adobe Dreamweaver.
For now, I was able to create a simple "Upload Excel" functionality which allows a user to upload the file and have it displayed perfectly on the web app. After this, we can do some statistics on the data.
The files will be on GitHub soon! I will also be deploying the project to the web at the end!
Week 8: Say Hello to Biographer!
Hello friends,
This week was very productive and we got a lot of things up and running.
First things first: the web app now has a proper name: Biographer! Get it? It's a "biology grapher," but "Biographer" is pronounced like the actual word. Currently the application is kind of ugly and "raw" but we'll worry about that later.
Last time, we were able to have the app display an Excel file. For core functionality, I got a simple dropdown to work:
But the main advancements I made this week had to do with user authorization. For starters, there is now a login screen:
This login feature also comes with a fully functional email system. If a user forgets their password, then the system will email them using the SMTP Gmail server: (I was lucky enough to get the "biographerhelp" email name)
Lastly, users have a profile which they can customize.
Tomorrow I'll integrate all the statistics scripts we wrote last month into the application. Next week we'll be close to finishing.
The end approaches!
Week 9: Biographer Updates
Hello friends,
This week Biographer got a whole load of updates. It is now very close to completion.
What's new:
Biographer can graph. Although currently I only have two stats scripts implemented, Linear Regression and Logistic Regression, the main logic is completed.
The Tools page is live! As promised, Biographer comes with four tools to make your lab life easier. The hemocytometer and notebook is directly coded into Biographer, while the timer and calculator redirect you to external sites. (It didn't make sense to me to reinvent the wheel)
The hemocytometer (cell counter) can count cells from an uploaded image using the TrackPy package. Here's a demonstration: (the formatting is a bit iffy but we'll fix that later):
And the Biographer Notebook, inspired by yatinbayya's "flask-notes-app" on GitHub, is a public notebook library that is password protected. This doubles as a Protocol Library, where users can upload their protocols for lab purposes and share them. (Passwords are optional!)
That's it for this week. The remaining to-do list to finish in Week 10:
Improving Uploads
Implementing more Stats
Beautification
It seems like with my changed GitHub username, you need a new link: https://github.com/samhuann/SeniorProject
Thank you!
Week 10: Biographer v1.0.0
Hello friends,
Can't you believe it? We're at the end! Last week, I fully wrapped up programming the core functionality and made it look a little bit better with Bootstrap. (Don't worry though, I'll still be making it look better before my final presentation)
Here's what we've got:
Transform
Transform data using a variety of functions. Also normalize/transpose for better data processing. (Like the new format?)
Statistical Tests
All sorts of statistical tests for your data.
Graphs
Let's do a two-way ANOVA:
Notice the new styling with Bootstrap. Although it looks slightly better, I'm not satisfied yet. Unfortunately I won't be able to show it through blog, but there is a presentation coming up and also you can see it through the deployed app.
Deployment
Using Amazon Web Services, I was able to deploy the app to the web so everybody could use it. However, due to blog deadlines, I wasn't able to fix some critical bugs, but they should be fixed in a day or two.
Biographer: biostatistics made easy
Well friends, this is it! I can proudly say I completed my senior project and now it's just time for some updates. Before final presentation I will be working on making it look even better and add better error handling.
Thank you!
-SH