Identifying biomarkers of anti-cancer drug synergy using multi-task learning

Aben, Nanne; Ruiter, de; Bosdriesz, Evert; Kim, Yonggyun; Bounova, Gergana; Dj, Vis; Lf, Wessels; Michaut, Magali

doi:10.1101/243568

Cited by 2 publications

(1 citation statement)

References 18 publications

(12 reference statements)

Supporting

Mentioning

Contrasting

Order By: Relevance

“…These strong AUPRCs demonstrate a fair trade-off between false positives and false negatives within our data ( Supp Fig 8 ). This is one of the highest reported accuracy measures for drug synergy prediction models based on any of these combination efficacy metrics 10, 29 .…”

Section: Resultsmentioning

confidence: 91%

Multi-task learning predicts drug combination synergy in cells and in the clinic

Gilvary

Elemento

2019

Preprint

View full text Add to dashboard Cite

18Combination therapies for various cancers have been shown to increase efficacy, lower toxicity, 19 and circumvent resistance. However, despite the promise of combinatorial therapies, the 20 biological mechanisms behind drug synergy have not been fully characterized, and the 21 systematic testing of all possible synergistic therapies is experimentally infeasible due to the 22 47Precision medicine has been instrumental in the creation of novel cancer therapeutics, however 48 many of these promising treatments have been limited by insufficient efficacy or acquired 49 resistance. Cancer cells have exhibited multiple types of resistance mechanisms, such as 50 redundant pathways 1 and pathway reactivation 2 , among others 3-5 . Combination therapies have 51 been proposed as a general strategy to combat therapeutic resistance, as well as increase overall 52 efficacy 6 . Such is the case with the combination of MEK and ERK inhibition to overcome 53 MAPK pathway reactivation 7 . Overall, rational drug combination therapies have the potential to 54 create long-lasting therapeutic strategies against cancer and perhaps many other diseases. 55 56While combination therapies have shown great potential, identifying effective drug pairings 57 presents a difficult challenge. Traditional methods for assessing drug efficacy of single drug 58 agents are not easily applied. With the number of approved or investigational drugs increasing 59 every year, the ability to pairwise test these agents across a wide breadth of disease models 60 becomes virtually impossible. This experimental infeasibility leads to the need for a 61 computational approach to predict drug synergy. 62 63Quantitative models have been introduced to predict effective drug combinations, however they 64 tend to be limited in scope, either confined to certain drug or cancer types 8 . A recent DREAM 65Challenge (dreamchallenges.org) 9 teamed up with AstraZeneca and called for models to predict 66 synergy across diverse drugs and cancer types 10, 11 . Despite >80 distinct computational methods 67 being submitted to harness biological knowledge and classify cells and drugs, with performance 68 matching the accuracy of biological replicates for many cases, there were still numerous drug 69 combinations that consistently performed poorly across all models. 70 71The problem is confounded as there currently is no agreed upon gold standard to measure drug 72 synergy of clinical relevance in vitro, introducing variability across method development and 73 misclassification of valuable drug combinations. There are many different metrics used to 74 measure drug synergy that are all based on models with differing underlying assumptions. 75Fourcquier and Guedi have presented a comprehensive and succinct description of many of these 76 models, how they differ and their practical limitations 12 . The most widely used of these methods 77include Bliss Independence 13, 14 , which assumes no interference/interaction between drugs, and 78Loewe Additivity 14, 15 , which is based on the d...

show abstract

Section: Resultsmentioning

confidence: 91%

Multi-task learning predicts drug combination synergy in cells and in the clinic

Gilvary

Elemento

2019

Preprint

View full text Add to dashboard Cite

show abstract

Machine learning and feature selection for drug response prediction in precision oncology applications

2018

View full text Add to dashboard Cite

In-depth modeling of the complex interplay among multiple omics data measured from cancer cell lines or patient tumors is providing new opportunities toward identification of tailored therapies for individual cancer patients. Supervised machine learning algorithms are increasingly being applied to the omics profiles as they enable integrative analyses among the high-dimensional data sets, as well as personalized predictions of therapy responses using multi-omics panels of response-predictive biomarkers identified through feature selection and cross-validation. However, technical variability and frequent missingness in input "big data" require the application of dedicated data preprocessing pipelines that often lead to some loss of information and compressed view of the biological signal. We describe here the state-of-the-art machine learning methods for anti-cancer drug response modeling and prediction and give our perspective on further opportunities to make better use of high-dimensional multi-omics profiles along with knowledge about cancer pathways targeted by anti-cancer compounds when predicting their phenotypic responses.

show abstract

Identifying biomarkers of anti-cancer drug synergy using multi-task learning

Cited by 2 publications

References 18 publications

Multi-task learning predicts drug combination synergy in cells and in the clinic

Multi-task learning predicts drug combination synergy in cells and in the clinic

Machine learning and feature selection for drug response prediction in precision oncology applications

Contact Info

Product

Resources

About