Bioinformatics as a tool for managing heterogeneous information, extracting knowledge about clinical/biomedical problems, from image diagnosis to identifying pathogenic genetic variants. In Struc. Bioinf. we focus on those problems happening at the molecular level. Most of them are related with the molecular view of disease, especially genetic disease. Examples: pathogenic variants in coding and non-coding sequence. These variants cause molecular malfunction; to understand this malfunction we need to understand function first and this means looking at structure. It is important to note that the information we have about structure will determine what we know about function. That is: quality determines usefulness. 

Relevant questions: (i) how to determine which information level we want; (ii) how do we extract 3D info; (iii) how to conclude from what we have? (seeking consistency between different sources of evidence), particularly in the context of biomedical applications.  

A. STRUCTURE: THE LOCUS OF FUNCTION 

1) PRINCIPLES OF STRUCTURAL BIOLOGY

1.1 Function at the sequence level. 

• Master Class
• Domains, disordered regions, repetitive regions. 
• Value and limitations. It is useful for a first assessment of the impact of protein variants; inadequate for deeper understanding. Not of value for drug design, nor for a mechanistic interpretation of disease

• Case Methods
• UniProt: the most important resource for general information on proteins. 
• Specific resources for functional domains, disordered, and repetitive regions. PFAM and SMART. 

1.2 The structure of macromolecules

• Master Class
• What do we mean by structure? Examples of 3D for different macromolecules, e.g., protein, RNA, DNA, and chromatin structure.
• Key relevant aspects of function that depend on structure: stability and interactions (the role of environment)

1.3 Relating structure to function. 

• Master Class

• The broad spectrum of protein function. Antibody, hemoglobin, a-galactosidase, BRCA1, etc. Moonlighting proteins.
• DNA: the structure of the genetic code
• Chromatin: packing is the key to compaction. But it also requires unpacking. 

• Case Methods.
• PDB: The experimental understanding of structure and the storage databases. PDB. EMDB. 

1.4 What parts of structure are relevant for function? It depends on the protein. How does bioinformatics allow us to know them? 

• Master Class. 
• Understanding structure I. (i) The hierarchy of protein structure: 1D, 2D, 3D, and quaternary. (ii) Core/Surface. (iii) The need to have a 3D structure in many problems.
• Understanding structure II. (i) The network of atomic interactions. Not all interactions are the same (ii) The energetics of 3D structures. Protein stability predictions (iii) Protein-complexes.
• Understanding structure III. Disordered proteins/regions. Resources

B. USING BIOINFORMATICS TO GENERATE STRUCTURAL INFORMATION

2) EXTRACTING INFORMATION FROM STRUCTURE: VISUAL EXPLORATION AND COMPUTATIONAL ANALYSES. 

• Master Class. 
• How to retrieve experimental structural information: (i) protein/gene name (uniprot); (ii) blast sequence
• Possible situations: is structure always available?
• Choose your visualization software: PyMol / VMD
• What to look at? The different representations we can choose: Calpha, main chain, all atoms, focus on local regions. 
• Clarifying your goal: understanding the contribution of a residue to the stability or its interaction with other molecules, define the amount of work
• Tools: Arpeggio, String. 
• Be careful with what you analyze: some structures are not meaningful biologically, e.g. crystal contacts, etc. Working with the monomer is the safest option, reading the original papers is the next step, checking specialized websites like Eldorado (for NMR) or biological complexes at the EBI.

• Case Methods: 
• PYMOL: Representing protein sequence variants

3) GENERATING STRUCTURAL INFORMATION: STRUCTURE PREDICTION

3.1 From sequence to structure: a key problem. 

• Master Class. 
• How sequence determines 3D. Why prediction is such a difficult problem?
• Alternatives: the most fundamental relationship in molecular biology: conservation vs structure/stability 
• Another fundamental principle: the conservation of structure beyond sequence identity

• Case Methods: 
• Homology. Sequence alignment and comparison. 
• Building and understanding MSA: generalizing sequence comparison to many sequences. A difficult problem: NP-complete. Main tools: clustal, T-Coffee, Muscle, etc. Representing and editing MSA. 

• Case Methods: 
• Structure comparison. Going beyond sequence similarity: Aligning different versions of the same structure. Finding how similar are different structures. Structure comparison: DALI. Structure classification: CATH, SCOP.

3.2 Homology/comparative modeling. 

• Master Class
• Predicting the structure of close homologs
• The comparative modeling pipeline
• Quality checks: PROSA, program listings, etc
• Online modeling: SwissProt
• Predicting the structure of remote homologs/paralogs: threading
• The fold space
• Threading: principles and tools.

• Master Class
• Ab initio structure prediction. AlphaFold: the new paradigm. 

3.3. Disorder predictions. 

• Master Class. 
• Disordered state
• Predicting disorder

4) APPLYING STRUCTURAL BIOINFORMATICS IN BIOMEDICINE

5) APPLYING MACHINE LEARNING TO UNDERSTAND DISEASE

• Master Class. 
• A mild introduction to artificial intelligence and machine learning. 

• Case Methods. 
• Building a machine learning model: four main steps
• Retrieving protein sequence variants. Clinvar. HGMD. Mapping variants to our protein: what information do we need to know. 
• Generating pathogenicity predictions for our variants. Polyphen-2, SIFT, PON-P2, CADD, REVEL, EVE, etc. 
• Predicting functional impact vs. clinical phenotype.

Clases magistrales: 15 hours 

Métodos del Caso: 15 hours

Prácticas: 0 hours