Einfuehrung in der Bioinformatik

Donnerstag 27. November 2008, EMBL Heidelberg

Aidan Budd and Evangelos Pafilis

Presentation

Demonsration - BLAST

BLAST webserver at EBI - we'll run it using one of the smaller sequence databases (UniProtKB/SwissProt) using the SRC_HUMAN sequence.

Note: The following demonstration was carried out on an OSX Macintosh, using the Firefox browser. While much will be similar using a different operating system and browser, there will certainly also be differences (for example, on OSX the Safari browser has quite a lot of problems recognising some of the files available for downloading from this page, and with displaying full information from the Ensembl website).

Demonstration - Hemoglobin

There is a bewildering diversity of bioinformatic data and tools - there will be corners (or even huge swathes) of the field that even a professional bioinformatician will know little or nothing about.

Thus, when using bioinformatic tools, it is important to focus addressing very specific questions - to avoid becoming overwhelmed by the multitude of opportunities and choices available.

With this in mind, this demonstration begins by focusing on a very specific question:

What effect does the mutation that causes sickle-cell anemia have on the 3D structure of haemoglobin?

We begin by identifying the UniProt (the major protein primary sequence database) record for human hemoglobin involved in sickle-cell anemia

We query using the EBI's EB-eye search tool with the terms "sickle-cell anemia"

NOTE: Searching bioinformatics databases usually matches multiple records that the user has to choose between to find the one(s) of interest

Following the "Protein Sequence" link we are led to the UniProt record for the protein, providing information about MANY aspects of the structure and function of the protein:

Taxonomy of the organisms (human)
Links to primary literature describing features of the protein
DNA sequence of the gene
3D structure of the protein
Phosphorylation state of the protein
Protein sequence of the gene
etc.

NOTE: Cross-references between biological databases make finding additional information about the protein/entities much easier and quicker. Almost all bioinformatic databases contain such links

From here we get the primary amino acid sequence of the protein (following the link that says "FASTA")
We can also find the mutation in the protein that causes sickle-cell anemia (searching the page for the word "sickle")
The UNFORMATED version of this data looks like this

NOTE: The unformated version of the data is relatively difficult for humans to interpret (it's designed to be relatively easy to process by computers) - therefore a key component of bioinformatics focuses on presenting the data in a way that is accessible to human users.

The UniProt record (or the EB-eye search) link to information about the 3D structures of the protein in the PDBsum database

with wildtype sequence/structure Link direct to PDB file
with the sickle-cell mutation Link direct to PDB file
NOTE - to acces the PDB file, follow the links to MSD or RCSB on the right side of the screen
(Looking at the PDB files directly using a text editor we can see what the raw data looks like)

NOTE: This example shows even more clearly why we need visualisation tools for bioinformatic data!

We can look a the structures of the wildtype and sickle-cell form of haemoglobin beta from human by loading them both into PyMOL

NOTE - loading the structures into PyMOL may be tricky. You might try:

drag and drop the files onto the PyMOL program icon
change the names of the files to ones expected by PyMOL e.g. ending in ".pdb"
Use the command prompt in the program to load the file e.g. "load /Users/budd/Desktop/pdb1gzx_hemoglobin_wt.ent.txt"

To begin to compare the structures we need to change the representations somewhat
Without placing the two structures directly on top of each other it's difficult to see the differences. Therefore we will...

Align the two structures using the CE webserver, comparing both B chains:

1gzx:B
2hbs:b

NOTE: CE, PyMOL, and EB-eye are examples of another key component of bioinformatics - TOOLS (as opposed to pure data) developed to help explore and analyse the basic bioinformatic data. One way of deconstructing bioinformatics is to divide it into "data" and "tools".

We can edit the results file (removing the "ENDMDL" and "MODEL 2" lines) and view the alignment in PyMOL

NOTE: All the tools and data we use here are FREE - with that in mind, it's not surprising that they are often not particularly easy to use. It took me at least 15 minutes to work out how to change the file to be able to view it properly in PyMOL. Problems like this are typical for bioinformatic tools.

We can also carry out some editing of the appearance of the structures in PyMOL and load this edited file directly into PyMOL load a pre-edited file into PyMOL to show the differences between the structures of the two chains

Note - again, you may have problems loading the edited file into PyMOL - in this case, try:

changing the name to end ".pse"
loading the file from the command prompt e.g. "load /Users/budd/Desktop/ceAlignmentHemoglobin_a_wt_b_sickleCell.pse"

How would you describe the differences between the two structures?

We have now answered our original question!

Mining the internet to obtain this range of information about a protein involves (as you've seen) using and understanding many different resources. Tools such as EB-eye help integrating many such resources (again, as we've seen) - however other tools have been develop that make the initial stages of this process (gaining an overview of what's know about a protein) in many cases very easy (and very attractively presented.)

REFLECT is one such tool and has an impresively wide range of diferent applications - we'll see here how it can help extracting relevant biological information from different types of text.

Returning to the hemoglobin 3D structure - this might be a useful starting point for many other questions e.g.:

What is the difference in the structure of human haemoglobin alpha and beta (providing insight into the evolution of the protein family)? CE alignment PyMOL Session
What is the effect on haemoglobin beta structure of other mutations (we can see a list of naturaly occuring variations in the UniProt record for the protein)?

Or using completely different structures:

Short functional peptide and protein disorder FHA domain intereating with phosphothreonine peptide PyMOL Session

Much of the proteome is DISORDERED i.e. has no globular structure
Disordered sequence is VITAL for many important functions (especially signalling e.g. phosphorylation - cancer, development, etc.)
This NMR structure shows an exmple of a disordered peptide

DNA complexed with HMGB protein

Can they read the DNA sequence by looking at the structure of the
Does the binding of the protein change the structure of the DNA (could include
Which regions of the structure do they expect to be be more/less flexible? (Show the movie)
Comparison with DNA in a non-double-helix conformation - what is the difference with how the bases intereact with each other?

Returning to haemoglobin as an example for other kinds of questions:

Where are the haemoglobin genes in the human genome (i.e. which chromosomes)?

Go to the Ensembl genome browser and query with HBB_HUMAN to see where it is located

NOTE: HBB_HUMAN is (sort of) an accession number for this protein used by UniProt (it's actually the entry's "name" - the primary accession number is P68871) - knowing what it is makes it much easier to find information relating to the protein in other databases. Associating information with an accession number is a very common feature of bioinformatic databases

Search within the gene page for "Family" - we can follow this link to information about where other the members of the family are located (all on chromosome 11)

Are they randomly distributed through the genome or are they clustered in one area?

Do the different haemoglobin genes have similar or different numbers of introns/exons? Do they have one or more transcripts for each locus?

Follow the links to each of the genes from the Ensembl family page and count!

How many copies of the gene are there in humans? In other mammals? When did the genes duplicate?

Query TreeFam with the accession number - this should lead to TF333268 - from here we can see when the duplication events occured

NOTE: TF333268 is the accession number for this family in the TreeFam database

How are the protein sequences of the members of the family different i.e. how have they changed during evolution?

From TreeFam choose "clean.aln" and click "Go"
Download the alignment and view with ClustalX locally

NOTE: ClustalX (as PyMOL previously) is a tool that we need to have installed directly on the computer we are working on - it is termed a "local" tool. This is in contrast to tools such as CE that are "web" or "remote" tools. The local/remote destinction is another way of describing features of different bioinformatic tools.

This shows which mutations have occured within the family during evolution

Is the pattern of substituion the same at all positions?
If there is a difference, which parts of the 3D structure are the fast/slow changing residues?
Are there other changes in the sequences? (Yes - insertions/deletions)
Do insertions/deletions occur preferentially in particular parts of the protein sequence/structure?
Which parts of the structure have slower/faster changes? etc.

How have the DNA sequences changed during evolution?

Choose "clean.nucl.mfa" and click "Go".
Download the alignment and view with ClustalX

This illustrates how mutation patterns vary between different positions in the alignment

Are there any obvious patterns/tendencies in how different positions in the sequence change?

Yes - it's due to the redundancy in the three-letter genetic code

Taking a short piece of DNA coding sequence, can they deduce the protein sequence it codes for?

This is the first few bases of the mRNA sequence for the HBB_HUMAN gene which tranlates to give the following peptide sequence

This can be done automatically using transeq from the EMBOSS package at EBI

There are many other directions one could try and go in developing tasks/projects/packages for school classroms:

Understanding bioinformatic tools (e.g. seqeunce alignment)

Sequence alignment is CENTRAL to ALMOST ALL sequence analysis bioinformatic tools. The aim of calculating an alignment is to place equivalent residues from different sequences in the same column.

For example, we can align two haemoglobin sequences HBA_HUMAN and HBB_HUMAN using Smith-Waterman alignments at EBI using EMBOSS or Blast2Sequences at the NCBI

To understand how these tools work, students could try and do this themselves without using these tools e.g. using paper and scissors or editing a simple text file.

There are many other such tools/tasks that could be investigated in this way e.g. matching linear motif regular expressions in ELM

Specific systems/data of interest

As mentioned, there are MANY MANY different datatypes, databases, and tools available. Some of the following websites provide lists of these, perhaps some of them would be particularly interesting for school students.

As an example, PhenoBank contains data on the effect of all C. elegans genes for their role in the first two rounds of mitotic cell division via a genome-wide RNAi screening and time-lapse video microscopy of the early embryo e.g. wild-type video

Some websites to visit that provide lists of many different bioinformatic tools and resources:

Cross-over with Informatics

An obvious area of interdisciplinary cross-over is with informatics. One could envisange many different tasks for computational students that would incorporate biological data e.g.:

Designing their own visualisation tools

3D structure e.g. PyMOL
text/html formating of database records e.g. as with UniProt where this raw data becomes this marked up HTML
interaction networks e.g. search with "HBB_HUMAN" in STITCH
...

Parsing useful information from large biological datasets

Implementing bionformatic algorithms

pairwise alignment
regular-expression mapping
mass-spec data interpretation
...

Another obvious cross-over field is mathematics.

Conclusion

We've looked here at numerous examples of questions that can be addressed using bioinformatic tools, and possible ideas for how such tools/ideas/approaches might be incorporated into the classroom. In summary here are two points I'd particularly like to emphasise:

All the data and tools used above is FREE for to access, download, and work with/analyse
Bioinformatic data and tools cover a very wide range of different aspects of biology - thus there may be many parts of the biology sylabus where it could be included

Novel Approaches for Automated Information Collection So far you have seen that drawing biology-related conclusions involves querying many different resources to collect the relevant pieces of knowledge. This is a tedious and time-consuming process that prevents researchers from spending their time on other research activities.

Clicking on the following button will cause the current page to be opened in a new window.

Have any protein names been highlighted within the page?
What happens if somebody clicks on the highlighted names?

Tools such as Reflect aim at automating the task of collecting relevant biological information by presenting the users with useful summaries about the proteins and small chemicals mentioned in a piece of text. Now vist the wikipedia entry of your proteins or small chemical molecule of your choice e.g. the prion protein, insulin, or Prozac

Copy and paste the URL of these wikipedia pages into the input field of Reflect. You should see similar marking up of protein names as before.

The same thing can be done with many different URLs e.g.:

Thus, without specialised bioinformatics skills or without prior knowledge of biological and chemical databases users can now get access to essential pieces of knowledge and easily realise, for example, how theoretical concepts such as a 3D structure materialise in practice.

Zuereck zur Gibson Team Ausbildungseiten