Irene Vlassi Pandi and Andrew Papanastasiou
Tue 02 Apr 2019, 11:00 - 12:00
IF 4.31/4.33

If you have a question about this talk, please contact: Gareth Beedham (gbeedham)

Irene Vlassi Pandi

Probabilistic Typing for TypeScript

As the need for properly specified and verified software emerges, the problem of automatically inferring rich type annotations for code becomes more compelling. Βy transcribing a type inference procedure as an optimization numerical problem we are enabled to combine hard and soft constraints to suggest types. As an instantiation of this setting, we consider the process of generating a TypeScript declaration file from a JavaScript library. We propose to combine into a single optimization problem: naming conventions for types learned by a character level LSTM model and an augmented static analysis expressed by using fuzzy logic. Given the popularity of JavaScript, improving the type inference process of its typed superset, TypeScript, should contribute towards a situation where the programmer efficiently achieves the best of both the dynamically and statically-typed worlds. This is joint work with Andy Gordon, Charles Sutton and Earl Barr (UCL).


Andrew Papanastasiou

Dissecting the importance of sequencing depth in scRNA-seq experiements

Single-cell RNA-sequencing provides biologists with a powerful tool to probe transcriptome heterogeneity of cell populations. Recent advances in scRNA-seq technologies have allowed the huge scaling-up of the number of cells that can be profiled in a single experiment. In this talk, using a simple toy model as well as empirical analysis of real data, we illustrate the impact that sequencing depth -- an experimental design parameter whose importance is often overlooked -- has on the power of generated scRNA-seq datasets to disentangle cell-subpopulations.