Cell Line Literature Mining
Background¶
As an extension of the standard "data resource" model of Progenetix, together in
a collaboration with a group from the ZHAW
we have implemented "knowledge exploration" entry points for cell line genomics.
Based on work from the INODE project the
ZHAW team has parsed collections of scientific publications with relation to
CCL analyses for the semantic identification of functional annotations, e.g.
the contextual association of genes or genomic regions.
Implementation¶
Based on a list of CCL-related publications (e.g. from the studies containing CNV
data in progenetix or those referenced in the cellosaurus entries) we utilize subject-predicate-object triple extraction from text, and entity-ontology linking
via a graph database. Relationships extracted in a context that references a given
cell line arer then represented for some selected categories. Here, following the
overall Progenetix paradigm of whoole genome (CNV) profile representation for a start
we focus on objects with genomic "mappability", especially genes and cytobands.
Fuzziness of cell line <-> object associations
The listing of a given object (e.g. gene) on a cell line page is based on the identification of the gene id in the referencing publication. However, such an association may be circumstantial (e.g. the gene being identified as target in a list of analyzed cell line samples).
User Guide¶
- Navigate to
Cell Line Listings - Use search bar to find the cell line of interest
- You will be redirected to cell line page where cell line metadata and variant information is displayed.
- By scrolling down, you can find the section Literature Derived Contextual Information.
- There, all gene results are listed. Gene of interest can be visualised on the CNV frequency plot by clicking on the gene.