Clare Bates Congdon

Common features of microRNA target prediction tools

The human genome encodes for over 1800 microRNAs (miRNAs), which are short non-coding RNA molecules that function to regulate gene expression post-transcriptionally. Due to the potential for one miRNA to target multiple gene transcripts, miRNAs are recognized as a major mechanism to regulate gene expression and mRNA translation. Computational prediction of miRNA targets is a critical initial step in identifying miRNA:mRNA target interactions for experimental validation. The available tools for miRNA target prediction encompass a range of different computational approaches, from the modeling of physical interactions to the incorporation of machine learning. This review provides an overview of the major computational approaches to miRNA target prediction. Our discussion highlights three tools for their ease of use, reliance on relatively updated versions of miRBase, and range of capabilities, and these are DIANA-microT-CDS, miRanda-mirSVR, and TargetScan. In comparison across all miRNA target prediction tools, four main aspects of the miRNA:mRNA target interaction emerge as common features on which most target prediction is based: seed match, conservation, free energy, and site accessibility. This review explains these features and identifies how they are incorporated into currently available target prediction tools. MiRNA target prediction is a dynamic field with increasing attention on development of new analysis tools. This review attempts to provide a comprehensive assessment of these tools in a manner that is accessible across disciplines. Understanding the basis of these prediction methodologies will aid in user selection of the appropriate tools and interpretation of the tool output.

Bioinformatic prediction of arthropod/nematode-like peptides in non-arthropod, non-nematode members of the Ecdysozoa

The Onychophora, Priapulida and Tardigrada, along with the Arthropoda, Nematoda and several other small phyla, form the superphylum Ecdysozoa. Numerous peptidomic studies have been undertaken for both the arthropods and nematodes, resulting in the identification of many peptides from each group. In contrast, little is known about the peptides used as paracrines/hormones by species from the other ecdysozoan taxa. Here, transcriptome mining and bioinformatic peptide prediction were used to identify peptides in members of the Onychophora, Priapulida and Tardigrada, the only non-arthropod, non-nematode members of the Ecdysozoa for which there are publicly accessible expressed sequence tags (ESTs). The extant ESTs for each phylum were queried using 106 arthropod/nematode peptide precursors. Transcripts encoding calcitonin-like diuretic hormone and pigment-dispersing hormone (PDH) were identified for the onychophoran Peripatopsis sedgwicki, with transcripts encoding C-type allatostatin (C-AST) and FMRFamide-like peptide identified for the priapulid Priapulus caudatus. For the Tardigrada, transcripts encoding members of the A-type allatostatin, C-AST, insect kinin, orcokinin, PDH and tachykinin-related peptide families were identified, all but one from Hypsibius dujardini (the exception being a Milnesium tardigradum orcokinin-encoding transcript). The proteins deduced from these ESTs resulted in the prediction of 48 novel peptides, six onychophoran, eight priapulid and 34 tardigrade, which are the first described from these phyla.

General video game playing

One of the grand challenges of AI is to create general intelligence: an agent that can excel at many tasks, not just one. In the area of games, this has given rise to the challenge of General Game Playing (GGP). In GGP, the game (typically a turn-taking board game) is defined declaratively in terms of the logic of the game (what happens when a move is made, how the scoring system works, how the winner is declared, and so on). The AI player then has to work out how to play the game and how to win. In this work, we seek to extend the idea of General Game Playing into the realm of video games, thus forming the area of General Video Game Playing (GVGP). In GVGP, computational agents will be asked to play video games that they have not seen before. At the minimum, the agent will be given the current state of the world and told what actions are applicable. Every game tick the agent will have to decide on its action, and the state will be updated, taking into account the actions of the other agents in the game and the game physics. We envisage running a competition based on GVGP playing, using arcadestyle (e.g. similar to Atari 2600) games as our starting point. These games are rich enough to be a formidable challenge to a GVGP agent, without introducing unnecessary complexity. The competition that we envisage could have a number of tracks, based on the form of the state (frame buffer or object model) and whether or not a forward model of action execution is available. We propose that the existing Physical Travelling Salesman (PTSP) software could be extended for our purposes and that a variety of GVGP games could be created in this framework by AI and Games students and other developers. Beyond this, we envisage the development of a Video Game Description Language (VGDL) as a way of concisely specifying video games. For the competition, we see this as being an interesting challenge in terms of deliberative search, machine learning and transfer of existing knowledge into new domains.

Preliminary Results for GAMI: A Genetic Algorithms Approach to Motif Inference

We have developed GAMI, an approach to motif inference that uses a genetic algorithms search and is designed specifically to work with divergent species and possibly long nucleotide sequences. The system design reduces the size of the search space as compared to typical window-location approaches for motif inference. This paper describes the motivation and system design for GAMI, discusses how we have designed the search space and compares this to the search space of other approaches, and presents initial results with data from the literature and from novel tasks. GAMI is able to find a host of putative conserved patterns; possible approaches for validating the utility of the conserved regions are discussed.

REALM: A rule-based evolutionary computation agent that learns to play Mario

REALM is a rule-based evolutionary computation agent for playing a modified version of Super Mario Bros. according to the rules stipulated in the Mario AI Competition held in the 2010 IEEE Symposium on Computational Intelligence and Games. Two alternate representations for the REALM rule sets are reported here, in both hand-coded and learned versions. Results indicate that the second version, with an abstracted action set, tends to perform better overall, but the first version shows a steeper learning curve. In both cases, learning quickly surpasses the hand-coded rule sets.

CARMEL Versus FLAKEY A Comparison of Two Winners

■ The University of Michigan’s CARMEL and SRI International’s FLAKEY were the first- and secondplace finishers, respectively, at the 1992 Robot Competition sponsored by the American Association for Artificial Intelligence. The two teams used vastly different approaches in the design of their robots. Many of these differences were for technical reasons, although time constraints, financial resources, and long-term research objectives also played a part. This article gives a technical comparison of CARMEL and FLAKEY, focusing on design issues that were not directly reflected in the scoring criteria.

Agent Smith: Towards an evolutionary rule-based agent for interactive dynamic games

The goal of this project is to develop an agent to play the first-person shooter game Unreal Tournament 2004 [1], a fast-paced and dynamic environment that demands that the agent must be capable of making decisions quickly. An additional goal of this project is to explore evolutionary computation as a means for learning the rule sets used to control the game-playing agent. The agents behavior is controlled by a rule-based system, which looks at multiple high-level conditions, such as whether the agent is weak, and determines a single high-level action, such as to head for the nearest known healing source. Using an evolutionary computation approach, in which the behavior is evolved over a number of generations, the agent learns increasingly better strategies for its environment. Through the work in this project, we are exploring several research questions, including the development of successful vocabulary of high-level conditions and actions for the rule set, the challenges of using the evolutionary process to hone a rule set, and the effects of using some expert knowledge in combination with the evolutionary process.

The use of robots in the undergraduate curriculum: experience reports

Using the robot as a metaphor for assisting students in understanding problem solving in general, the algorithmic process, and the relationship between algorithms and computing agents is not new. While simulated robot environments have existed for many years (e.g. Karel the Robot[3]) it is only recently that the technology for inexpensively supplying undergraduates with real robots has become available. Lego Mindstorms, MIT Handyboards, the Rug Warrior, and others are examples of such systems. Programmable in familiar languages, including C, Ada, and Java, these systems allow for the creative exploration of important computer science concepts. Representing a variety of institution types the panelists will discuss their experiences in using hands-on robot-based projects for illustrating various important computer science concepts.

Integrated mobile-robot design-Winning the AAAI 1992 robot competition

The Carmel project (computer-aided robotics for maintenance, emergency, and life support) which won the AAAI 1992 Robot Competition, is discussed. Carmels design philosophy and architecture, obstacle avoidance, global path planning, vision sensing, landmark triangulation, and supervisory planning system are described. The Carmel project shows that mobile robots can perform carefully chosen tasks reliably and efficiently, although this requires extensive integration of components and a solid engineering effort. >

Phylogenetic trees using evolutionary search: initial progress in extending Gaphyl to work with genetic data

Gaphyl is an application of evolutionary algorithms to phylogenetics, an approach used by biologists to investigate evolutionary relationships among organisms. For datasets larger than 20-30 species, exhaustive search is not practical in this domain. Gaphyl uses an evolutionary search mechanism to search the space of possible phylogenetic trees, in an attempt to find the most plausible evolutionary hypotheses, while typical phylogenetic software packages use heuristic search methods. In previous work, Gaphyl has been shown to be a promising approach for searching for phylogenetic trees using data with binary attributes and Wagner parsimony to evaluate the trees. In the work reported here, Gaphyl is extended to work with genetic data. Initial results with this extension further suggest that evolutionary search is a promising approach for phylogenetic work.