AI Antibiotic Drug Discovery

A recently published article, about using AI to help discover a new class of chemical antibiotics, is likely one of the first of a wave of applications of artificial intelligence techniques to solving scientific problems.

The Hard Fork podcast has an excellent interview with one of the researchers involved in the study. They do a great job describing the problem (the search for antibiotics to deal with drug resistant bacteria) and overviewing how the AI was combined with physical experiments to do the research.

Python Setup on Different Operating Systems (2021)

Because students bring their own laptops and devices I’ve had some trouble in the past getting everyone running python with the packages I typically use (mainly vpython). This is a quick reference about how to get up and running on Windows, OSX, and Linux.

Figure 1. Test sphere (from a Linux installation).

Linux

These should be pretty identical. Ultimately, students may want to learn how to set up virtual environments (e.g. venv), however, for our purposes python3 should already be installed.

Using a Terminal window we’ll install pip:

sudo apt install python3-pip

and use pip to install vpython (which should install numpy as a dependency).

pip3 install vpython

and finally matplotlib:

pip3 install -U matplotlib

Test the installation by running python3 and making a sphere. Run:

python3

In the interpreter enter the two lines:

from vpython import *
sphere()

A new browser tab should pop up with a sphere (Figure 1).

OSX

OSX should be very similar to the Linux installation. You may have to install the latest version of python from https://www.python.org/downloads/, but you can check to see if python3 is installed by typing into a Terminal window:

python3

which will start the interpreter if python3 is installed.

Windows

Download and install the latest version of python (3.9 at the moment) from https://www.python.org/downloads/ . I then followed the instructions on Installing Packages for Windows.

The following commands will be run using the Command Prompt application:

First check that python is installed (and which version it is):

py --version

which should show something like (if it does not then something probably went wrong with the installation):

Python 3.9.7

Now install and upgrade pip (and setuptools and wheel):

py -m pip install --upgrade pip setuptools wheel

Install vpython (which includes numpy)

py -m pip install vpython

And matplotlib:

py -m pip install matplotlib

Test by opening the IDLE App, and at the interpreter prompt typing the two lines:

from vpython import *
sphere()

and you should see Figure 1 pop up in a new browser tab.

Bright and Shiny Things: Programming with LED’s

Teaching programming using the LED light strips is going much better than expected. I tried it with the 9th grade Algebra class during our weekly programming session using a set of coding lessons I put together. I went so well that though we started by having everyone (about 10 kids) share two LED strips, by the end of the year I had three students from that class build their own.

Student built LED strip.
Student’s LED strip on a sword. The battery can power it for at least 15 minutes.

The coding lessons are still a work in progress, but it has them learn the basics by running some of the test programs, then then explore sequences using for loops. There are a lot of directions to branch off after the for loops. I’ve had some of my Algebra II students make static patterns using linear and exponential functions, while a couple of the kids in my programming class used different functions to make dynamic lighting patterns; our hydroponic system (see here and here) now has a neat LED indicator that runs different sequences depending on if the pumps are running or not.

Some of the students who built their LED strips in the Makerspace posted about their projects: LED Thingy and LED Light Strip Project. The process (rpi-led-strip) is not too hard but required them to be able to do a little physical computing (with Raspberry Pi’s), use ssh and terminal commands (terminal instructions), and then run and write python programs.

Raspberry Pi that controls one of the LED strips from a student’s project.

Since the setup uses the same GitHub repository (rpi-led-strip) it’s also easy to update some of our existing projects like the Wall Anchor.

Wall anchor project.

I am amazed at how much the students have engaged with what are, ultimately, very simple systems (a Raspberry Pi and a strip of 20 lights), and I’m really excited to see where it takes us.

Python Tutorial (on jobtensor)

In talking to alumni and their peers who have recently started computer science related careers, I have been encouraged to continue using python as the primary language for introducing students to programming.

I was recently pointed towards a very good Introduction to Python on the jobtensor website. It has a clean design and uses iPython Shells on the webpage so you can run some basic code on the site itself.

From Simple Equations to Complex Behavior

Another excellent video from Veritasium. Starts with the logistic equation and through a series of very clear examples gets to the relationship between growth rate (r) and equilibrium population. He does go into how this graph relates to the Mandelbrot set, but the rotating graphs are a little tricky to follow. However, the discussion of the practical applications of chaos theory (at 10:35) is really nice as well.

Note:

The logistic equation:

 x_{n+1} = r x_n (1-x_n)

gives the new population (xₙ₊₁) for a given growth rate (r) and the old population (xₙ).

This makes for a fairly nice and easy programming assignment.

HTML Animations using SVG

A basic animation test using the HTML SVG element instead of the CANVAS. This one uses the svg.js library. This library has it’s own, built in animation function that is used to make the red rectangle move back and forth (1 second period).

The green circle is moved using the requestAnimationFrame function with some coding to make it bounce back and forth.

<!DOCTYPE html>
<html lang="en" dir="ltr">
  <head>
    <script src='./svg.js/svg.min.js'></script>
    <meta charset="utf-8">
    <title></title>
  </head>
  <body>
    <div id="svgPlace"></div>
  </body>

  <script type="text/javascript">
    var draw = SVG().addTo('#svgPlace').size(300,400);
    var line = draw.line([[20,20],[200,20]]).stroke({width: 1, color:'#f00'});

    //Metronome red rectangle.
    rect = draw.rect(50,100).move(200,20).fill('#f00');
    rect.radius(10);
    dir = 'left';

    setInterval(animate, 1000);
    function animate(){
      if (dir === 'left'){
        px = 20;
        py = 20;
        dir = "right";
      }
      else {
        px = 200;
        py = 20;
        dir = "left";
      }
      rect.animate({
        duration: 1000,
        when: 'now'
      }).move(px,py);
    }

    //Green circle
    circ = draw.circle(20).fill('#0f0').move(250,10);
    window.requestAnimationFrame(animateCirc);
    cx = 250;
    cy = 20;
    dy = 1;

    function animateCirc(){
      cy+=dy;
      circ.move(cx,cy);
      window.requestAnimationFrame(animateCirc);
      if (cy > 100 ){
        cy = 100;
        dy = -dy;
      } else if (cy < 20) {
        cy = 20;
        dy = -dy;
      }
    }

  </script>
</html>

Making Animations with HTML Canvas

Based off of Mozilla’s nice introduction to basic animations on the web, I made as simple of an animation as I could using the Canvas element in HTML. It just contains a static, red rectangle, and a green sphere that slowly moves across the canvas. This should serve as a simple introduction for my students who want to make online games.

<!DOCTYPE html>
<html lang="en" dir="ltr">
  <head>
    <meta charset="utf-8">
    <title></title>
  </head>

  <body>
    <div id="gameBoard"></div>

  </body>

  <script type="text/javascript">

    canvas_width = 400;
    canvas_height = 300;

    //create canvas
    cvs = document.createElement("CANVAS");
    cvs.height = canvas_height;
    cvs.width = canvas_width;
    document.getElementById("gameBoard").append(cvs);
    ctx = cvs.getContext("2d");
    cx = 10;
    cy = 20;

    window.requestAnimationFrame(animate);

    function animate(){
      //clear the canvas
      ctx.clearRect(0, 0, canvas_width, canvas_height); // clear canvas

      //set location of the circle
      cx += .2;
      cy += .1;
      
      //recreate the canvas (every time)
      ctx.fillStyle = '#ff0000';
      rect = ctx.fillRect(0, 0, 20, 100);

      ctx.beginPath();
      circ = ctx.arc(cx, cy, 10, 0, 2 * Math.PI);
      //ctx.stroke();
      ctx.fillStyle = '#00ff00';
      ctx.fill();

      window.requestAnimationFrame(animate);

    }

  </script>
</html>

This uses plain HTML and Javascript, with no additional libraries.

Environments for Teaching Python in Middle and High School

I’ve been using python as the primary programming language for all of my classes, from middle to high school. However, with a variety of different machines and operating systems–macs, PC’s, OSX, Windows, ChromeOS–it has been bit of a pain getting everyone up and running and on the same page. It has been especially problematic because I like to use NumPy and, especially, the VPython module because of its nice, easy 3d visualizations.

I’ve tried Anaconda, which works great on some machines but not so much on others. One student with a Mac had to restart the kernel every time he ran a program with vpython, while others had trouble getting it to work properly at all. I do like the way it allows you to put everything into a single Jupyter Notebook, which works nicely for teaching-length programming assignments, but becomes cumbersome for longer projects. Joel Grus goes into detail about why he does not like Notebooks for teaching; much of which I agree with. I do like the Spyder IDE a lot, but I’ve run into some computers that have problems with it (constant crashes).

So, I’ve been using Glowscript a lot with the lower grades. It’s online, so you only need a browser, and its made for vpython, so you can do a lot of the numerical work with it. However, because it’s online you can’t, understandably, get it to write output files to your computer–you have to copy and paste from its output. I’m also not sure how to import user-created modules, which further limits its utility for higher-level classes.

I’ve recently run into repl.it, and I’m having a student try it out in my Numerical Methods class. It looks quite promising as it seems to be able to manage files well and we’ve tried some basic numerical programs using numpy and matplotlib. However, I’ve not got it running vPython yet (although it’s come close).

(repl.it update): One of our other teachers has been trying repl.it with the middle school class and really likes it.

I’ve also recently discovered MakeCode. I’m looking into it for the lower grades, down to 6th grade or even earlier, because it lets you write programs for Minecraft, which is very popular with a remarkably broad age group. It lets you work in blocks, python, or JavaScript. From my poking around, the first couple of things it teaches you to do in Minecraft are write statements (to generate a chicken) and the loops (to generate a lot of chickens). I’ve asked one of my 9th grade, Minecraft players to investigate, so I should have some results to report soon.

At the moment, for my most advanced class, we’re taking a whatever-works approach. We’re learning about finite differences and using matplotlib to graph the output. I have one student, mentioned above, using repl.it, and another one using Jupyter Notebook on a Mac. A different Mac user is invoking IDLE from the command line because they prefer the IDLE IDE, but couldn’t get numpy to install properly with the python 3.9 version they’d installed earlier–the command-line IDLE actually goes to their Anaconda installation. I have one student, on a Windows machine, for whom everything just works and are using their regular IDLE installation. We spent the entire class period getting everyone up and running with something, but at least now everyone has at least one environment they can use. I’m curious to see what the kids who have multiple options are going to go with.