Application talk
Abstract
Despite concerted efforts by the biotech, pharma industries and the research community (e.g [1]), the cost of therapy development continues to rise and many diseases lack efficient treatment. A key driver for this stubbornly upward trend is the necessity to develop personalized drug therapies, taking into account the patient’s’ genotype, that target increasingly complex disease mechanisms. The ability to infer biologically relevant relations from complex and cross domain data analysis could direct research towards the most promising targets for therapy. The Euretos Target Identification is an interactive application to find novel target candidates for the development of chemical or biologic treatments of human disease. The application is build on top of the Euretos Knowledge Platform, an innovative life sciences information platform used by some of the world’s largest pharma & biotech companies, academic hospitals and research institutes [2]. The platform connects more than 4,5 million phenotypic, genetic, proteomic, metabolic and chemical concepts from over 60 data & publication sources in a single ultra-performance environment [3]. Euretos has solved the data integration problem by designing a semantic model for integrating heterogeneous data. The semantic approach is based on the nanopublication model, a minimal model for describing a subject-predicate-object assertion together with attribution and provenance metadata [4]. Using the Target Identification application, researchers can minimize the risk of downstream target failure by 1) Creating an exhaustive list of novel target candidates based on differential expression and target involvement in pathways, expression regulation and protein complexes, 2) Analyzing target ranking based on combined efficacy, druggability and safety scoring, 3) Evaluating the underlying scientific evidence for each potential target, 4) Assessing target efficacy even when no direct literature references exist. Key functional capabilities crucial to target discovery include: 1) Select differential expression conditions: The basis of the Euretos Target Identification application is differential expression. Many experiments with varying conditions are available for any disease. The minimum log2fold differential expression threshold is determined by the user. 2) Expand target list with associated targets: The initial target list can be expanded by adding related targets in associated pathways, protein complexes and expression regulation. The optional targets can be filtered on relating to up or down regulated targets only. 3) Simultaneously compare targets on all aspects of efficacy, safety and druggability: Each target from the target candidates list is scored on expected efficacy, druggability and safety. The highest ranked targets are on top as primary candidates to review. 4) Assess druggability: Key aspects of the druggability of the target are available for assessment: all known compounds with relevant mode of action (i.e. agonist for down regulated targets) and affinity levels (at least 100nM), the Human Protein Atlas drug candidate assessment [5] and whether the target is already an U.S. Food and Drug Administration approved drug target. 5) Assess safety: Key aspects of the safe use of the target are available for assessment: baseline expression to determine expression specificity, the number of genetic interactions with the target (and genes involved) and the number of diseases the target is associated with (and the specific diseases involved). 6) Assess efficacy: Key aspects of the efficacy of the target are available for assessment: differential expression, the functional association with the disease, the number of disease annotations for known gene variants, the number of publications in which the target and the disease co-occur and the number of physiological processes associated with the target that are regarded as hallmarks of cancer (for neoplastic targets only). 7) Review efficacy for novel targets where no literature (yet) exists: In many cases the suggested targets are entirely novel for the disease and only linked through differential expression. The Target Identification application provides all indirect relations for the target – disease association including an assessment of the strength of biological interaction. This provides a unique and powerful way to assess efficacy immediately. The prediction algorithm is inspired by the concept profile technology [6]. Euretos has adapted this technology to fit a graph-based semantic network. A highlight is the predicted therapeutic target Aurora Kinase 1 (AURKA) for breast cancer. The Target Identification application suggest AURKA as a top 3% target candidate. A recent article [7] has shown that the AURKA gene plays a role in the development of breast cancer via a transcription interaction in the nucleus. Most cancer drugs are based on kinase inhibition, thus targeting AURKA opens up an alternative way to develop a drug therapy that is not kinase dependent. The life sciences deal with inherently complex questions that need a current answer. Extracting the right knowledge from heterogeneous data sources is non-trivial. The scope and breadth of the Euretos Knowledge Platform are unparalleled. Perhaps more importantly, the Euretos Knowledge Platform is unique in enabling powerful embedded (predictive) analytics that are too resource intensive to run in traditional environments. This enables very powerful applications such as Target Identification where we combine many aspects of efficacy, safety and druggability simultaneously to suggest promising drug targets. At every stage we provide the researcher access to the underlying references to assess the underlying evidence.
References:
[1] Guney E, Menche J, Vidal M, Barábasi AL. Network-based in silico drug efficacy screening. Nat Commun. 2016;7:10331
[2] http://euretos.com/references
[3] http://euretos.com/images/oct2015/Data201601c.pdf
[4] Mons B, et al. The value of data. Nat Genet. 2011;43(4):281-3
[5] http://www.proteinatlas.org/humanproteome/druggable
[6] Hettne KM, et al. The Implicitome: A Resource for Rationalizing Gene-Disease Associations. PLoS One. 2016;11(2):e0149621
[7] Zheng F, et al.Nuclear AURKA acquires kinase-independent transactivating function to enhance breast cancer stem cell phenotype. Nat Commun. 2016; 7: 10180.
Authors
Aram Krol, Euretos, Netherlands
Arie Baak, Euretos, Netherlands
Onno Becker Hof, Euretos, Netherlands
Kristina Hettne, Leiden University Medical Center, Netherlands