Semantic network for bioactive compound discovery from scientific literature

1 . A method for automated therapy discovery comprising:

accessing a corpus of scientific publications;

compiling a population of semantic concepts represented in the corpus of scientific publications into a vector space model based on:

proximity of semantic concepts within individual scientific publications, in the corpus of scientific publications; and

frequency of semantic concepts across the corpus of scientific publications;

deriving domains of a set of chemical and biological concepts in the vector space model based on proximity to domain descriptors in the vector space model;

deriving association scores between connected chemical and biological concepts, in the set of chemical and biological concepts, based on proximity in the vector space model;

deriving action characteristics between connected chemical and biological concepts, in the set of chemical and biological concepts, based on action descriptors in the vector space model;

generating a semantic network comprising:

a set of nodes representing the set of chemical and biological concepts labeled with domains; and

connections between nodes storing association scores and action characteristics;

receiving a query for a target concept and a target domain at a research portal;

generating a set of hypotheses by:

isolating a set of edges, in the semantic network, between a target node representing the target concept and a subset of nodes labeled with the target domain;

for each edge in the set of edges in the semantic network:

identifying a subset of chemical and biological concepts along the edge in the semantic network; and

generating a hypothesis, in a set of hypotheses, for a direction and a magnitude of an effect of the subset of chemical and biological concepts on the target concept based on association scores and action characteristics stored in connections along the edge; and

returning the set of hypotheses, ranked by magnitude of effect, to the research portal.

2 . The method of claim 1 , further comprising, for a first node, in the subset of nodes, representing a first concept in the target domain:

identifying a first population of edges, in the semantic network, connecting the target node and the first node;

for each edge in the first population of edges:

calculating an intermediate association score, in a first set of intermediate association scores, based on association scores and directions contained in connections between nodes along the edge in the semantic network; and

calculating a first composite association score for the first concept, in the target domain and represented by the first node, based on a first combination of the first set of intermediate association scores.

3 . The method of claim 1 , wherein generating the set of hypotheses comprises:

identifying the target node in the semantic network;

defining a radius limit for a distance from the target node to nodes in the target domain; and

identifying the subset of nodes, in the semantic network, in the target domain and within the radius limit of the target node.

4 . The method of claim 1 , wherein isolating the set of edges in the semantic network comprises isolating the set of edges in the semantic network connecting the target node to nodes, labeled with the target domain, separated by fewer than a threshold quantity of intermediate nodes in the semantic network.

5 . The method of claim 1 , further comprising:

projecting sets of edges, in the semantic network, between the target node and the subset of nodes onto a virtual surface to generate a visualization of a region of the semantic network representing connections between the target concept and the target domain;

labeling edges, represented in the visualization, within concepts extracted from nodes between the target node and the subset of nodes in the semantic network; and

rendering the visualization within the research portal for the user.

6 . The method of claim 1 :

wherein receiving the query for the target concept and the target domain at the research portal comprises:

receiving selection for the target concept comprising a target compound at the research portal; and

receiving selection for the target domain comprising disease concepts at the research portal;

wherein generating the set of hypotheses comprises:

scanning the semantic network for disease concepts;

isolating a set of disease nodes, in the semantic network, nearest the target concept representing the target compound;

for each disease node in the set of disease nodes:

isolating a first set of edges coupling the disease node to the target concept;

calculating a composite association score between the disease node and the target concept based on a combination of association scores and directions contained in the set of edges;

isolating a secondary set of edges coupling the disease node to the target concept; and

calculating a secondary composite association score between the disease node and the target concept based on a secondary combination of association scores and directions contained in the secondary set of edges; and

wherein returning the set of hypotheses to the research portal comprises:

compiling a first list of disease concepts contained in the set of disease nodes, ranked by secondary composite association score; and

presenting the first list of disease concepts labeled as side effects of the target compound, within the research portal to the user.

7 . The method of claim 1 :

wherein receiving the query for the target concept and the target domain at the research portal comprises:

receiving selection for the target concept comprising a target disease at the research portal; and

receiving selection for the target domain comprising bioactive compounds at the research portal;

wherein generating the set of hypotheses comprises:

isolating the set of edges, in the semantic network, between the target node representing the target disease and the subset of nodes labeled as bioactive compounds; and

for each node, in the subset of nodes, labeled with a bioactive compound concept:

isolating a first set of edges coupling the node to the target concept;

calculating a composite association score between the bioactive compound concept and the target disease based on a combination of association scores and directions contained in the first set of edges;

isolating a second set of edges coupling the node to a nearest secondary node, in the semantic network, labeled with a taste quality; and

calculating a taste association score for the bioactive compound concept based on a second combination of association scores contained in the second set of edges; and

wherein returning the set of hypotheses to the research portal comprises:

compiling a first list of the set of bioactive compound concepts ranked by composite association score; and

rendering the first list of the set of bioactive compound concepts, labeled with taste qualities and taste association scores, within the research portal to the user.

8 . The method of claim 1 :

wherein receiving the query for the target concept and the target domain at the research portal comprises:

receiving selection for the target concept comprising a target disease at the research portal; and

receiving selection for the target domain comprising compound concepts at the research portal;

wherein generating the set of hypotheses comprises:

isolating the set of edges, in the semantic network, between the target node representing the target disease and the subset of nodes labeled as compound concepts; and

for each node, in the subset of nodes, labeled with a compound concept:

isolating a first set of edges coupling the node to the target concept; and

calculating a composite association score between the compound concept and the target concept based on a combination of association scores and directions contained in the first set of edges; and

wherein returning the set of hypotheses to the research portal comprises:

compiling a first list of compound concepts, ranked by composite association score; and

rendering the first list of compound concepts for presentation within the research portal to the user.

9 . The method of claim 8 , further comprising, in response to selection of a high association strength filter from the research portal:

accessing a first definition of high association strength;

identifying a first subset of compound concepts, from the first list of compound concepts, exhibiting high association strength based on the first definition of high association strength;

compiling the first subset of compound concepts into a second list of compound concepts ranked by composite association score and labeled with high association strength; and

rendering the second list of compound concepts within the research portal for the user.

10 . The method of claim 1 :

wherein receiving the query for the target concept and the target domain at the research portal comprises:

receiving selection for the target concept comprising a target disease at the research portal; and

receiving selection for the target domain comprising bacteria concepts at the research portal;

wherein generating the set of hypotheses comprises:

isolating the set of edges, in the semantic network, between the target node representing the target disease and the subset of nodes labeled as bacteria concepts; and

for each node, in the subset of nodes, labeled with a bacterium concept:

isolating a first set of edges coupling the node to the target concept; and

calculating a composite association score between the bacterium concept and the target disease based on a combination of association scores and directions contained in the set of edges; and

wherein returning the set of hypotheses to the research portal comprises:

compiling a first list of bacteria concepts, ranked by composite association score; and

rendering the first list of bacteria concepts for presentation within the research portal to the user.

11 . The method of claim 1 , wherein deriving action characteristics between connected chemical and biological concepts, in the set of chemical and biological concepts, based on action descriptors in the vector space model comprises:

deriving action characteristics representing directions of correlations between connected chemical and biological concepts based on the presence of directional keywords between chemical and biological concepts within individual scientific publications of the corpus of scientific publications.

12 . The method of claim 11 , wherein deriving action characteristics comprises deriving action characteristics representing directions of correlations between connected chemical and biological concepts based on the presence of directional keywords, directional keywords is selected from the group consisting essentially of:

upregulates;

downregulates;

catalyzes;

inhibits;

starts;

stops;

causes;

prevents;

promotes;

demotes;

grows;

kills;

induces; and

reduces.

13 . The method of claim 1 , wherein generating the set of hypotheses comprises:

identifying a first sequence of nodes along a first edge, in the set of edges, connecting a first concept in the target domain to the target concept in the semantic network;

extracting a first sequence of chemical and biological concepts from the first sequence of nodes;

extracting a first set of association scores and a first set of action characteristics stored in a first series of connections between the first sequence of nodes along the first edge;

deriving a first direction of a first effect of the first concept on the target concept based a first combination of the first set of action characteristics;

deriving a first magnitude of the first effect, in the first direction, based on a second combination of the first set of association scores; and

compiling the first direction and the first magnitude into a first hypothesis for the first effect of the first concept on the target concept.

14 . The method of claim 13 , further comprising:

initializing a first natural language script;

inserting a first sequence of words, representing the first sequence of chemical and biological concepts, into the first natural language script;

transforming the first set of action characteristics into a second set of words;

inserting the second set of words, interposed between the first sequence of words, in the first natural language script according to a language model;

populating the first natural language script with the first magnitude and the first effect; and

rendering the first natural language script, describing the first hypothesis in natural language, within the research portal.

15 . A method for automated therapy discovery comprising:

accessing a corpus of scientific publications;

compiling a population of semantic concepts represented in the corpus of scientific publications into a vector space model;

deriving domains of a set of chemical and biological concepts in the vector space model based on proximity to domain descriptors in the vector space model;

deriving association scores between connected chemical and biological concepts, in the set of chemical and biological concepts, based on proximity in the vector space model;

deriving action characteristics between connected chemical and biological concepts, in the set of chemical and biological concepts, based on action descriptors in the vector space model;

generating a semantic network;

receiving a query for a target concept and a target domain at a research portal;

identifying a target node representing the target concept and a subset of nodes labeled with the target domain in the semantic network;

generating a set of hypotheses by:

identifying a subset of biological and chemical concepts in the target domain nearest the target concept;

for each concept in the subset of biological and chemical concepts:

isolating a set of edges coupling the concept to the target concept;

calculating a composite association score between the concept and the target concept based on a combination of association scores and directions contained in the set of edges; and

generating a hypothesis, in a set of hypotheses, for a direction and a magnitude of an effect of the concept on the target concept based on association scores and action characteristics stored in connections along the set of edges; and

rendering a first list of concepts, ranked by association score and linked to the set of hypotheses, for presentation within the research portal for the user.

16 . The method of claim 15 , further comprising, in response to selection of a first concept in the first list of biological and chemical concepts:

rendering a first hypothesis, in the set of hypotheses, within the research portal; and

rendering a first natural language script, describing the first hypothesis in natural language, within the research portal.

17 . The method of claim 15 , further comprising:

projecting sets of edges, in the semantic network, between the target node and the subset of nodes onto a virtual surface to generate a visualization of a region of the semantic network representing connections between the target concept and the target domain;

labeling edges, represented in the visualization, with concepts extracted from nodes between the target node and the subset of nodes in the semantic network; and

rendering the visualization within the research portal for the user.

18 . The method of claim 15 , further comprising, in response to selection for a direct documentation status filter at the research portal:

extracting a first subset of biological and chemical concepts from the list of biological and chemical concepts, exhibiting a direct documentation status;

compiling the first subset of biological and chemical concepts into a second list of biological and chemical concepts according to the direct documentation status filter; and

rendering the second list of biological and chemical concepts, labeled with direct documentation statuses, within the research portal for the user to review.

19 . The method of claim 15 , further comprising:

in response to selection of a first publication date threshold at the research portal:

identifying a first subset of scientific publications in the corpus of scientific publications exceeding the first publication date threshold; and

extracting a first cluster of identifiers from the semantic network corresponding to the first subset of scientific publications;

in response to selection of a second publication date threshold from the research portal:

identifying a second subset of scientific publications in the corpus of scientific publications falling below the second publication date threshold; and

extracting a second cluster of identifiers from the semantic network corresponding to the second subset of scientific publications;

generating a second list of concepts ranked by publication date and labeled with the first cluster of identifiers and the second cluster of identifiers; and

presenting the second list of concepts, labeled with the first cluster of identifiers and the second cluster of identifiers, within the research portal for the user.

20 . A method for automated therapy discovery comprising:

accessing a corpus of scientific publications;

compiling a population of semantic concepts represented in the corpus of scientific publications into a vector space model based on:

proximity of semantic concepts within individual scientific publications, in the corpus of scientific publications; and

frequency of semantic concepts across the corpus of scientific publications;

deriving domains of a set of concepts in the vector space model based on proximity to domain descriptors in the vector space model;

deriving association scores between connected concepts, in the set of concepts, based on proximity in the vector space model;

deriving action characteristics between connected concepts, in the set of concepts, based on action descriptors in the vector space model;

generating a semantic network;

receiving a query for a target concept and a target domain at a research portal;

isolating a set of edges, in the semantic network, between a target node representing the target concept and a subset of nodes labeled with the target domain;

identifying a subset of concepts along each edge of the set of edges in the semantic network;

generating a hypothesis, in a set of hypotheses, for a direction and a magnitude of an effect of the subset of concepts on the target concept based on association scores and action characteristics stored in connections along each edge of the set of edges; and

returning the set of hypotheses, ranked by magnitude of effect, to the research portal.