What To Cover In A Short Talk On Bioinformatics

Question

[Farhat]

I am giving a 1.5 hour talk on Bioinformatics in which I would like to cover the following topics.

• Various schemes for constructing phylogenetic trees: salient features, pros and cons.
• Available tools for analysing sequence data.
• Querying Public Sequence databases.

The talk is aimed at Biology PhD students with generally no programming background. What tools or websites would you cover given the time constraints? Anything that you think a biologist should know but I am missing in the above short list?

[Giovanni M Dall'Olio]

can you give more information? Is this the only training that these students are going to have on bioinformatics? Are them from the same discipline, and do they work in the same field?

[Farhat]

For most of them this may be the only training they get. I intend to give pointers to them to follow up deeper on a particular topic if they feel like it.

Answer 1

[Giovanni M Dall'Olio]

You can take inspiration from the Software Carpentry for Bioinformatician project. SC is a project to create an introductory documentation for bioinformaticians and researchers wishing to enter this field, it has been around for almost 10 years and is a highly recommended resource for any new researcher.

They also have many slides already available.

You can also look at the Programming4Scientist blog, which unfortunately seems to have not been updated lately, but has a good archive nevertheless.

Update:

let me see, an outline for a talk could be:

• introduce Margaret Dayhoff as the founder of bioinformatics. Explain how she created the PAM matrices to solve the problem of comparing two sequences, and to quantify the 'similarity' between two proteins.
• explain public databases, at least the NCBI ones and Uniprot.
• explain how computational tools can save time and money in a research project. You can make the example of drug modeling that makes easier to test new drug candidates. Most old school researches (and their students) think of bioinformatics as not a science, and that using computational tools is not research. You should try to convince them of the opposite, with good examples.
• explain how a software must be carefully tested before being considered complete. Wet scientists will like that, because they like designing tests for their experiments. You can make the analogy that, as a wet scientist has to design controls and markers for a western blot, a bioinformatician has to design the right tests to prove that her/his program works perfectly, without testing too much or not enough
• some wet biologists are afraid to share their data with someone else or to release it publicly. They think that since they have created the data, they should do all the analysis by themselves. Explain that, even if they have this right, a computational scientist with a good experience in analyzing data can make a better analysis, in a much shorter time. Moreover, it is important to compare the data produced by many independent researchers, convince that if they share their data, somebody else can compare their results with others and get a better picture of the hidden elephant.

[Giovanni M Dall'Olio]

I have had an idea. You could take the classic figure of 'blind scientists analyzing an elephant' and add the case of bioinformaticians. In this case, the work of a bioinformatician is to collect together the data from different experiments made by wet biology researchers; for example, in http://www.whattofix.com/blog/archives/2008/05/peace-for-pachy.php , a bioinformatician would take the results from two or more researchers ('it is a wall' and 'it is a rope') and make some conclusion ('it is a wall with a rope to move it').

Answer 2

[Pierre Lindenbaum]

I would use the NCBI as a central point for your course.

• Search for a sequence using GenBank ("what is an accession ? etc...")
• Blast this sequence (quickly explain "what is blast ?")
• ask Blast to make a phylogenetic output
• Search for the literature about this sequence

Answer 3

[Biomed]

I would make sure I have enough time to cover Galaxy. I find it very useful to introduce a bench scientist to Galaxy for simple and common data analysis tasks that have no scientific value to me but is desperately needed by the bench scientist to get some answers.

Answer 4

[Zhidkov]

Hi,

If you perform your talk to biologists about phylogeny I think you should start with short review about alignment:

Beginning:

1. What is alignment and why it is MP complete problem (ClustalW, T-Coffe, Muscle etc.)
2. Why prefer amino acid sequence (when possible) and not DNA
3. What are the differences between alignment, BLAT and BLAST

Phylogeny:

1. What is phylogenetic tree stands for (remind the alignment problem)
2. Types of phylogentic trees (NJ, MaxPars etc) and why they differ (why it will be not so smart to try to construct a Maximum Parsimony tree of chromosome 1 full sequence from 800 individuals which one can download from 1000 genome project
3. What is bootstrap stands for.

4. Tools for building phylogenetic trees:

   • PHYLIP
   • PAUP
   • MEGA
   • EMBOSS

Databases and usage:

• NCBI
• UCSC
• Ensembl
• Galaxy

Best,
Ilia