Week 7: Starting out

    This is my first week in the TRAIN program, meaning half of my work was orientation and reviewing the new project. I did well on the safety test, which is good because I plan on working with bacteria! Our project is to create a machine learning algorithm that can determine the type of bacteria from 8 different samples we have based on their characteristics. We will initially use Microsoft Azure or Google's Teachable Machine to input our data and develop a proof of concept. Then, we will attempt to code the algorithm and create our own machine-learning program ourselves. It will be difficult, but with the 2 other people on the project, this is totally doable. 

    After finishing orientation, safety tests, and lab tours, I started working on creating my first 8 samples of each bacteria we were using. The bacteria we are using are:

EA Enterobacter aerogenes  -bacilli

EC Escherichia coli -bacilli

PA Pseudomonas aeruginosa  -bacilli

PS Providencia stuartii -bacilli

SM Serratia marcescens -bacilli

 

BS Bacillus subtilis +bacilli

CX Corynebacteria xerosis +bacilli

BS Bacillus cereus +bacilli

 

SE Staphylococcus epidermidis +cocci

SA Staphylococcus aureus  +cocci

EF Enterococcus faecalis +cocci

I waited 24 hours for the samples to develop, and half turned out quite well. On the samples that were good to use, there were distinguishable single colonies of bacteria that we could take our first pictures of. However, the other half needed to be redone as my streaks were heavy-handed. Not bad for my first time working with bacteria, though! After redoing the samples, Joshua and I used a dissecting microscope. However, the pictures needed to be more detailed to show distinct characteristics of the colony other than it being a sphere (at least for our SA sample). Therefore, we switched to a regular microscope, where the pictures turned out much better! We will likely use that microscope for the rest of our samples. On this day, we did 3 samples. One was E. Coli, which had a very irregular shape, but the other two bacteria we scanned (SA and EF). These two looked very similar, reddish-brown spheres with a darker center. Therefore, we plan to make 3 different algorithms for each class so that the program can distinguish between class types instead of looking for characteristics in all 8 samples simultaneously. This will be our plan until we figure out how to take more detailed photos. 

    For now, my next few weeks will look similar as we will need to take many, many photos to create a reliable algorithm. We made significant progress to start, though!

    E.Coli

 EF

   SA 

Sampling bacteria!