Deep sequencing of short capped RNAs reveals novel families of noncoding RNAs.

In eukaryotes, capped RNAs include long transcripts such as messenger RNAs and long noncoding RNAs, as well as shorter transcripts such as spliceosomal RNAs, small nucleolar RNAs, and enhancer RNAs. Long capped transcripts can be profiled using cap analysis gene expression (CAGE) sequencing and other methods. Here, we describe a sequencing library preparation protocol for short capped RNAs, apply it to a differentiation time course of the human cell line THP-1, and systematically compare the landscape of short capped RNAs to that of long capped RNAs. Transcription initiation peaks associated with genes in the sense direction have a strong preference to produce either long or short capped RNAs, with one out of six peaks detected in the short capped RNA libraries only. Gene-associated short capped RNAs have highly specific 3' ends, typically overlapping splice sites. Enhancers also preferentially generate either short or long capped RNAs, with 10% of enhancers observed in the short capped RNA libraries only. Enhancers producing either short or long capped RNAs show enrichment for GWAS-associated disease SNPs. We conclude that deep sequencing of short capped RNAs reveals new families of noncoding RNAs and elucidates the diversity of transcripts generated at known and novel promoters and enhancers.

Temporal encoding in deep reinforcement learning agents.

Neuroscientists have observed both cells in the brain that fire at specific points in time, known as "time cells", and cells whose activity steadily increases or decreases over time, known as "ramping cells". It is speculated that time and ramping cells support temporal computations in the brain and carry mnemonic information. However, due to the limitations in animal experiments, it is difficult to determine how these cells really contribute to behavior. Here, we show that time cells and ramping cells naturally emerge in the recurrent neural networks of deep reinforcement learning models performing simulated interval timing and working memory tasks, which have learned to estimate expected rewards in the future. We show that these cells do indeed carry information about time and items stored in working memory, but they contribute to behavior in large part by providing a dynamic representation on which policy can be computed. Moreover, the information that they do carry depends on both the task demands and the variables provided to the models. Our results suggest that time cells and ramping cells could contribute to temporal and mnemonic calculations, but the way in which they do so may be complex and unintuitive to human observers.