Background: Recently high-throughput technologies have been massively used alongside clinical tests to study various types of cancer. Data generated in such large-scale studies are heterogeneous, of different types and formats. With lack of effective integration strategies novel models are necessary for efficient and operative data integration, where both clinical and molecular information can be effectively joined for storage, access and ease of use. Such models, combined with machine learning methods for accurate prediction of survival time in cancer studies, can yield novel insights into disease development and lead to precise personalized therapies. Results: We developed an approach for intelligent data integration of two cancer datasets (breast cancer and neuroblastoma) − provided in the CAMDA 2018 ‘Cancer Data Integration Challenge’, and compared models for prediction of survival time. We developed a novel semantic network-based data integration framework that utilizes NoSQL databases, where we combined clinical and expression profile data, using both raw data records and external knowledge sources. Utilizing the integrated data we introduced Tumor Integrated Clinical Feature (TICF) − a new feature for accurate prediction of patient survival time. Finally, we applied and validated several machine learning models for survival time prediction. Conclusion: We developed a framework for semantic integration of clinical and omics data that can borrow information across multiple cancer studies. By linking data with external domain knowledge sources our approach facilitates enrichment of the studied data by discovery of internal relations. The proposed and validated machine learning models for survival time prediction yielded accurate results. Reviewers: This article was reviewed by Eran Elhaik, Wenzhong Xiao and Carlos Loucera.

Metagenomics studies, as well as genomics studies of polyploid species such as wheat, deal with the analysis of high variation data. Such data contain sequences from similar, but distinct genetic chains. This fact presents an obstacle to analysis and research. In particular, the detection of instrumentation errors during the digitalization of the sequences may be hindered, as they can be indistinguishable from the real biological variation inside the digital data. This can prevent the determination of the correct sequences, while at the same time make variant studies significantly more difficult. This paper details a collection of ML-based models used to distinguish a real variant from an erroneous one. The focus is on using this model directly, but experiments are also done in combination with other predictors that isolate a pool of error candidates.

For metagenomics datasets, datasets of complex polyploid genomes, and other high-variation genomics datasets, there are difficulties with the analysis, error detection and variant calling, stemming from the challenges of discerning sequencing errors from biological variation. Confirming base candidates with high frequency of occurrence is no longer a reliable measure because of the natural variation and the presence of rare bases. The paper discusses an approach to the application of machine learning models to classify bases into erroneous and rare variations after preselecting potential error candidates with a weighted frequency measure, which aims to focus on unexpected variations by using the inter-sequence pairwise similarity. Different similarity measures are used to account for different types of datasets. Four machine learning models are implemented and tested.

Summary This paper presents a study in the domain of semi-automated and fully-automated ontology mapping. A process for inferring additional cross-ontology links within the domain of anatomical ontologies is presented and evaluated on pairs from three model organisms. The results of experiments performed with various external knowledge sources and scoring schemes are discussed.