le (5m36s)

# Résultats de recherche

**703**

le (8m57s)

## 3.10. Gene prediction in eukaryotic genomes

If it is possible to have verygood predictions for bacterial genes, it's certainly not the caseyet for eukaryotic genomes. Eukaryotic cells have manydifferences in comparison to prokaryotic cells. You rememberthe existence of a nucleus and you also remember on one ofthe schemes in the first week that there are more structureswithin a eukaryotic cell. But the differences lie also inthe organization of the genomes. In eukaryotic genomes, the so-calledintergenic regions are very long. Intergenic regions are theregions which separate genes. A bacterial genome is very denseindeed, if you put your fingers somewhere on the genome, if itwas possible of course, it would be on the gene. If you do the ... Voir la vidéole (4m55s)

## 4.1. How to predict gene/protein functions?

Last week we have seen that annotating a genome means first locating the genes on the DNA sequences that is the genes, the region coding for proteins. But this is indeed the first step,the next very important step is to be able to predict thefunctions of the genes. That is more correctly, the function of the protein coded by the genes. How can we predict thisgene or function protein? It is essentially based on thefact that we will retrieve genes or protein for which the sequenceis similar and for which we know the function. So we will seehow we can measure and compute the similarity between DNA or protein ... Voir la vidéole (6m39s)

## 4.7. Alignment costs

We have seen how we can compute the cost of the path ending on the last node of our grid if we know the cost of the sub-path ending on the three adjacent nodes. It is time now to see more deeply why these costs are used to compute the cost in the last node. So again, we saw how we can compute the cost here of the path ending on that node if we know the cost of the sub-path ending on these three red nodes. Indeed, if we come from that node, the cost on that node will be the cost of that node plus the ... Voir la vidéole (4m12s)

## 4.6. A path is optimal if all its sub-paths are optimal

A sequence alignment between two sequences is a path in a grid. So that, an optimal sequence alignmentis an optimal path in the same grid. We'll see now that a property of this optimal path provides us with scanned lines for designing an optimization algorithm. The property is the following. A path which is optimal is made up of optimal sub-paths. To prove that, we can start byproving that if a path of length L is optimal then the path of length L minus one is also optimal. This can be proved quiteeasily ad arburdum. That is, you take the hypothesis that the path of length L is optimal ... Voir la vidéole (3m51s)

## 4.5. A sequence alignment as a path

Comparing two sequences and thenmeasuring their similarities is an optimization problem. Why? Because we have seen thatwe have to take into account substitution and deletion. During the alignment, the comparison of the two sequences, we haveto insert blank characters at a certain position in order tohave an optimal score that is we want the sequence to be themore similar as possible. So the problem is to find whereto locate the blank character. There are many solutions and wewant to find the best one, it is an optimization problem. How do we deal with thisoptimization problem? We will consider an alignmentbetween two sequences as a path in that kind of grid. Here ... Voir la vidéole (4m23s)

## 4.4. Aligning sequences is an optimization problem

We have seen a nice and a quitesimple solution for measuring the similarity between two sequences. It relied on the so-called hammingdistance that is counting the number of differencesbetween two sequences. But the real situation is a bitmore complex as we'll see now, it needs an adequatesolution and algorithm. Why is it a bit more complex? Let's have a look at thispair of two sequences. If we apply the hamming distance,compute the hamming between these two sequences,we find ten differences. OK. But you must remember thatmutation may be substitution, deletion and insertion. So if wetake into account the deletion and insertion, the situation isvery different in the case of these two sequences. ... Voir la vidéole (4m0s)

## 4.3. Measuring sequence similarity

So we understand why gene orprotein sequences may be similar. It's because they evolve togetherwith the species and they evolve in time, there aremodifications in the sequence and that the sequence may still besimilar, similar enough again to retrieve information on onesequence to transfer it to another sequence of interest. So thequestion now is how can we measure this similarity between twosequences for the moment. The first approach to similarityis a very simple one is to apply a distance which is calledhere the Editing System or the Hamming Distance.The idea is very basic. You would take two sequences likethese two sequences here and you look at the differences and youcount ... Voir la vidéole (4m30s)

## 4.2. Why gene/protein sequences may be similar?

Before measuring the similaritybetween the sequences, it's interesting to answer the question: why gene or protein sequences may be similar? It is indeed veryinteresting because the answer is related to the theory ofevolution which is due, as you all know, to Darwin. What Darwinsays is that species evolve in time and there is a creation ofnew species for existing ones. So there is an evolutionof species over time. He was a very thinking man, huh. This evolution can be also seenon the genomic sequences. Let's see this very small and partialtree of life and hypothetical tree of life. Here you have thespecies and you have this phenomenon of speciation giving ... Voir la vidéole (7m42s)