Hello everyone! I just completed my second to last week at CMU studying computational biology.
While I have enjoyed just about every minute of my time here, this week was particularly fun. I still spent some time in the lab doing biology stuff, but most of this week was spent coding one really interesting project that I can’t wait to tell you about. Before we get to that though, I want to update you all on the status of that bacteria that I might get my name on a technical paper for. Some of you may remember that I talked about this in my other blogs, but the short version is that some of the other students under Joshua Kangas cultivated a bacteria from river water samples that has yet to be entirely sequenced.Mentor Josh and I are going to sequence it and then he will write a paper if it’s new information which my name (as well as those of the other students involved) will be on. On Friday, I finished the DNA extraction portion of this process and if all went well then it will be sent off for sequencing tomorrow.
Now, that was an important step, but actually only took an afternoon to complete, so most of the remainder of my time was spent on that really interesting project that I mentioned above. Basically, I placed a bunch of cells on a “plate” (I’ll put in a picture of this for clarification) which has a number of “wells” and then I treated some of those wells with different doses of different drugs. Then I put that plate into the CX7 microscope and had it take images of all the different wells. Josh then tasked me with writing code that will convert all of those images into a TIFF format (because they were initially in some obscure biology format no one uses called a .C01) and creating a Convolutional Neural Network that could distinguish between the different drugs based only on the images. This was an exercise in creating a Deep Learning AI that sounds like the terminator but is really just complicated math. I’m not done with the project yet, but I’m closer than I would have expected and I’ve probably learned more from this than almost any other project of my life.
A Plate with 96 wells
This classifier itself is unlikely to be very useful in and of itself because, I am only going to be able to develop it for a week and Josh has undergraduates that have spent much more time making a better version of this, but the image conversion code may actually prove helpful and I cannot stress enough how big a learning opportunity it was for me. During this process Josh took a very hands off approach and allowed me to do everything without assistance (except for a fellow high school student named Eliot who I had worked with before and who was back in Pittsburgh on a 3 day flight layover). Eliot and I spent hours bouncing ideas off one another, banging our heads against the brick wall that is debugging, making more and more wild guesses as to why a particular problem was happening. and generally having a blast writing these programs. I even spent a lot of my free time working with him on this project which is why I don’t have anything particularly special to say about leisure activities this week.
Overall this has been one of the best weeks of my Pinternship, and, while I will be happy to be home again, I will be sad to leave this coming Saturday.
SO INTERESTING!!