AKS: Scientific Knowledge Navigation

The results of a literature analysis can sometimes be overwhelming. Powerful result visualization is needed to reduce complexity, allowing you to focus on the important features without getting lost in the details. AKS provides an effective and robust graphical interface that represents the relationships between the biological entities in a search, as a graph that is interactive and can be dynamically modified by the user. This allows you to visualize how a group of entities is related. For example, representation of gene networks or disease-chemical relationships, in order to investigate different classes of drugs.

Example
This example will show how can we see the most relevant information of thousands of documents in a small screen and how this perspective can derive in amazing new discoveries.

We start by loading three genes into AKS (see Fig. 1). This group of genes codifies the enzymes catalyzing the reactions in the biosynthesis of neurotransmitters (catecholamines) dopamine and norepinephrine.

We request for the cooccurrences among the enzymes (Fig. 2). Select every node and click on the right button. Click on the "Relate with graph" option and set up the parameters. The result is the graph displayed in Fig. 3. The edges represent the existence of documents where both of the connected genes cooccurr. Now, we want to plot the cooccurrences of this group of three genes with other genes in the biomedical knowledge. (Learn more about cooccurrences, see Extraction of biomedical concept relationship)

AKS provides different systems to produce graph layouts. Firstly, we try with the mode "random" and we see that the result is a graph excessively compact and unclear (Fig. 3). Now we try a more "intelligent" mode called "Organic - Classic". After setting a few parameters, we obtain a much more useful layout that shows a sort of peripheral subgroup and a central subgroup (Fig. 4).

The resulting graph contains 234 nodes (genes) and 343 edges (relationships among the genes) found in 3,841 documents.

We can move along this layout and manipulate manually its aspect. We made "drag and drop" with some of the subsets of genes and then we zoomed an area that seems interesting (Fig. 5, the zoomed area is the one inside the square). This enables to extract in a glance the genes that are exclusively connected to one of the first three genes (DBH) (Fig. 6).