Pickling in Python can be dangerous.
That’s where Quickle comes in — as long as you’re using Python 3.8 or later…
To Pickle, Or Not To Pickle? — Quickle!
On The Logic of GOing with Weisfeiler-Lehman
Recently, I was able to attend Martin Grohe’s talk on The Logic of Graph Neural Networks. Professor Grohe of RWTH Aachen University, is a titan of the fields of Logic and Complexity theory. Even so, he is modest about his achievements, and I was tickled when it was pointed out to me that the theorem he refers to as “a little complex”, one of his crowning achievements, involves a four-hundred page long book of a proof.
The theorem relates to the Weisfeiler-Lehmann (WL) algorithm, an algorithm for determining whether two graphs are equivalent (i.e. isomorphic). The algorithm has deep connections with combinatorics, complexity theory and first order logic. A system of logic that is remarkably similar to the relations present in ontologies such as the Gene Ontology (GO), which is commonly used to compare and predict protein function. Kernelised methods and other WL-based metrics present a new and possibly logically “complete” way to potentially compare the functions of proteins and infer their similarity.
Ten Simple Rules For Solving Any Problem
Welcome! Take three deep breaths, each time expel the air through your nose with force. Now you are ready for this adventure. Let us dive right in and reflect on the premise of this blog post.
I, personally, dislike the word “solve”. What does it even mean? And that is already the second time I have used that word. The word solve implies all kinds of nonsense, such as completion or the existence of a solution. Let us recast it as: new insights, positive reframing or simply “ah-ha!”. These “problems” can be anything too: emotional ones (external or internal), scientific or research ones, artistic ones, writing ones. If you feel like it, just call it a problem.
Whoever is writing this blog post, he certainly does talk a lot … We should really get going or we will run out of time. You know what, let us start again. Welcome to:
Continue readingPlotly for interactive 3D plotting
An recently wrote a post on how to use the seaborn library. I really like seaborn and use it a lot for 2D plots. However, recently I have been dealing with 3D data and have found plotly to be best. When used in a jupyter notebook, it allows you to easily generate 3D interactive plots. This is extremely useful to visualize structural data.
Ribosome occupancy profiles are conserved between structurally and evolutionarily related yeast domains
Shameless plug for any OPIG blog readers to take a look at our recent publication in Bioinformatics. Consider giving it a read if the below summary grabs your attention.
Many proteins are now known to fold during their synthesis through the process known as co-translational folding. Translation is an inherently non-equilibrium process – one consequence of this fact is that the speed of translation can radically influence the ability of proteins to fold and function. In this paper we compare ribosome occupancy profiles between related domains in yeast to test the hypothesis that evolutionarily related proteins with similar native folds should tend to have similar translation speed profiles to preserve efficient co-translational folding. We find strong evidence in support of this hypothesis at the level of individual protein domains and across a set of 664 pairs of related domains for which we are able to compute high-quality ribosome occupancy profiles.
To find out more, view the Advance Article at Bioinformatics.
Seaborn 101
Seaborn is a Python-based data visualization library, which is based on matplotlib (https://seaborn.pydata.org/) . I would like to share some guidance/code to get started with drawing plots using this library! I will be using the dataset ‘flights’ from Seaborn (https://github.com/mwaskom/seaborn-data) to highlight an example.
Continue readingCalculating symmeterised small molecule RMSDs using graph automorphisms in python with GEMMI and NetworkX
When a ring flips, how do we calculate RMSD?
This surprisingly simple question leads to a very interesting problem! If we take a benzene molecule, say, and rotate it 180 degrees, then we have the exact same molecule, but if we have a data structure in which our atoms are labelled, and we apply the same transformation to the atomic positions, the numbering does not reflect that symmetry. If we were then naively to calculate the RMSD it would be huge, despite the fact that the molecule is, chemically speaking, identical.
How can we make our RMSD calculations reflect these symmetries?
Continue readingTracking machine learning projects with Weights & Biases
Optimising machine learning models requires extensive comparison of architectures and hyperparameter combinations. There are many frameworks that make logging and visualising performance metrics across model runs easier. I recently started using Weights & Biases. In the following, I give a brief overview over some basic code snippets for your machine learning python code to get started with this tool.
Continue readingMiniproteins – small but mighty!
Proteins come in all shapes and sizes, ranging from thousands of amino acids in length to less than 20. However, smaller size does not correlate with reduced importance. Miniproteins, which are commonly defined as being less than 100 amino acids long, are receiving increased attention for their potential roles as pharmaceuticals. A recent paper by David Baker’s group put miniproteins into the spotlight, as the study authors were able to design miniproteins that bind the SARS-CoV-2 spike protein with as strong affinity as an antibody would – but in a tiny fraction of the size (Cao et al., 2020). These miniproteins are much cheaper to manufacture than antibodies (as they can be expressed in bacteria) and can be highly stable (with melting temperatures of >90º possible, meaning they can easily be stored at room temperature). The most promising miniprotein developed by the Baker group (LCB1) is currently undergoing testing to be used as a prophylactic nasal spray that provides protection against SARS-CoV-2 infection. These promising results – and the speed in which progress was made – brings the vast potential of miniproteins in healthcare to the fore.
Continue readingMaking Pretty Pictures with PyMOL
There’s few things I like more in our field than the opportunity to make a really nice image of a protein structure. Don’t judge me, but I’ve been known to spend the occasional evening in front of the TV with a cup of tea and PyMOL open in front of me! I’ve presented on the subject at a couple of our research group retreats, and have wanted to type it up into a blog post for a while – and this is the last opportunity I will have, since I will be leaving in just a few weeks time, after nearly eight years (!) as an OPIGlet. So, here goes – my tips and tricks for making pretty pictures with PyMOL!
Ray Tracing
set ray_trace_mode, number
I always ray trace my images to make them higher quality. It can take a while for large proteins, but it’s always worth it! My favourite setting is 1, but 3 can be fun to make things a bit more cartoon-ish.
You can also improve the quality of the image by increasing the ‘surface_quality’ and ‘cartoon_sampling’ settings.