Bioconductor is a successful project for the analysis and comprehension of high-throughput data. A key component of this success is the development of innovative statistical algorithms relevant to leading-edge bioinformatic data types. This proposal focuses on three aspects of algorithm development, to enable continued success of the project. The first aim is (1) to implement methodological and data structure support for signal recovery and inference across modern genomic assays. Signal recovery and inference are fundamental steps in the transformation of raw data into meaning summaries. The second aim is (2) to develop annotation and workflow support for genome biology and genomic medicine. This recognizes the importance of associating statistical insight with biological meaning. The final aim is (3) to provide the necessary infrastructure to support scalable computational analysis of genome-scale data. A key idea emerging from this work is the STAMP (scan, transform, amalgamate, model, pack) pattern for genomic analysis.