Bioinformatics tools to predict protein properties from sequence
10 September 2019
There are now well over 130 million protein sequences available from Uniprot, and for only a small fraction of these (about 2%) structural information is available from the Protein Data Bank. Uniprot provides a mix of manual and automatic functional and structural annotations for these sequences, but when working on a particular research question you might want to obtain more detailed insights into the possible behavior of the protein(s) you are interested in.
In this workshop, we address how to predict features of proteins from their sequence only, using Python scripts to access APIs of web services available online. We will introduce single-sequence based predictions of biophysical characteristics of proteins (e.g. DynaMine for backbone dynamics), methods that use multiple sequence alignments (e.g. I-TASSER), and methods that determine the difference between a protein and its mutant form (e.g. DEOGEN2). We will also show how the results of these predictions can be combined and analysed using a protein case study.
Basic knowledge of Python programming
Wim Vranken is research professor at the VUB in Brussels, and director of the (IB)2 VUB/ULB Interuniversity Institute of Bioinformatics in Brussels.
Niloofar Shirvanizadeh is a postdoctoral researcher at the Interuniversity Institute of Bioinformatics in Brussels. She has a Ph.D. in Biophysics and her work focuses on the structural bioinformatics and the use of protein structures in the field of (bio)medical life sciences.
Python set up
Bio2byte services (DynaMine, DisoMine, DEOGEN2, EFoldMine)
Other bioinformatics services (Jpred 3, ELM, I-TASSER, PROFEAT, TMHMM, STRING, SiteMap)