Adrian Caciula

Developing a multifunctional network laboratory for teaching and research

Adequate hands-on experience on networking and computing is becoming vital and necessary for students majored in programs such as Information Technology and Computer Science. However, building a hands-on experimental lab environment turns out to be challenging for many institutions, particularly for a new-born program or an urban university, due to space constraints, budget limitations, maintenance difficulty and so on. This paper presents and explores the design of a multifunctional lab environment for both teaching and research while taking the cost and space challenges into consideration. On one hand, the lab utilizes a number of industrial-strength Cisco routers, switches and wireless access points to setup a practical platform that allows students to configure and administrate their network design by applying the knowledge learned in the classroom. On the other hand, the lab is equipped with useful (and mostly free) software packages such as OPNET, Network Simulator (NS-2), Virtual PC, and CPLEX, which enable students to conduct various network modeling, simulation, optimization, and emulation together with their research projects. Accordingly, each iteration for the lab design and implementation as well as experiences learned and future integration with additional curriculum, are presented in this paper.

Transcriptome assembly and quantification from Ion Torrent RNA-Seq data.

BackgroundHigh throughput RNA sequencing (RNA-Seq) can generate whole transcriptome information at the single transcript level providing a powerful tool with multiple interrelated applications including transcriptome reconstruction and quantification. The sequences of novel transcripts can be reconstructed from deep RNA-Seq data, but this is computationally challenging due to sequencing errors, uneven coverage of expressed transcripts, and the need to distinguish between highly similar transcripts produced by alternative splicing. Another challenge in transcriptomic analysis comes from the ambiguities in mapping reads to transcripts.ResultsWe present MaLTA, a method for simultaneous transcriptome assembly and quantification from Ion Torrent RNA-Seq data. Our approach explores transcriptome structure and incorporates a maximum likelihood model into the assembly and quantification procedure. A new version of the IsoEM algorithm suitable for Ion Torrent RNA-Seq reads is used to accurately estimate transcript expression levels. The MaLTA-IsoEM tool is publicly available at: http://alan.cs.gsu.edu/NGS/?q=maltaConclusionsExperimental results on both synthetic and real datasets show that Ion Torrent RNA-Seq data can be successfully used for transcriptome analyses. Experimental results suggest increased transcriptome assembly and quantification accuracy of MaLTA-IsoEM solution compared to existing state-of-the-art approaches.

Improved transcriptome quantification and reconstruction from RNA-Seq reads using partial annotations

The paper addresses the problem of how to use RNA-Seq data for transcriptome reconstruction and quantification, as well as novel transcript discovery in partially annotated genomes. We present a novel annotation-guided general framework for transcriptome discovery, reconstruction and quantification in partially annotated genomes and compare it with existing annotation-guided and genome-guided transcriptome assembly methods. Our method, referred as Discovery and Reconstruction of Unannotated Transcripts (DRUT), can be used to enhance existing transcriptome assemblers, such as Cufflinks, as well as to accurately estimate the transcript frequencies. Empirical analysis on synthetic datasets confirms that Cufflinks enhanced by DRUT has superior quality of reconstruction and frequency estimation of transcripts.

RNA-Seq based discovery and reconstruction of unannotated transcripts in partially annotated genomes

We describe the problem of transcriptome reconstruction in partially annotated genomes from RNA-Seq that involves unannotated transcripts discovery and reconstruction in the context of incomplete annotation. In this paper we present an annotation-guided transcriptome reconstruction method and compare it with existing annotation-guided and genome-guided assemblers. Our Detection and Reconstruction of Unannotated Transcripts (DRUT) method is used to enhance existing transcriptome assemblers. Our method enables accurate detections of reads from unannotated transcripts and applying transcriptome assembler on subset of reads improve reconstruction quality. Empirical analyses on synthetic datasets show that enhancement of existing transcriptome assemblers with DRUT can increase number of assembled transcripts from RNA-Seq data.

TRIP: a method for novel transcript reconstruction from paired-end RNA-seq reads

Background Recent advances in DNA sequencing have made it possible to sequence the whole transcriptome by massively parallel sequencing, commonly referred as RNA-Seq. RNA-Seq is quickly becoming the technology of choice for transcriptome research and analyses. RNA-Seq allows to reduce the sequencing cost and significantly increase data throughput, but it is computationally challenging to use such RNA-Seq data for reconstructing of full length transcripts and accurately estimate their abundances across all cell types. A number of recent works have addressed the problem of transcriptome reconstruction from RNA-Seq reads. These methods fall into three categories: genomeguided, genome-independent and annotation-guided.

Deterministic regression algorithm for transcriptome frequency estimation

We present a deterministic version of our novel Monte-Carlo Regression based method MCReg [1] for transcriptome quantification from RNA-Seq reads. Experiments on simulated and real datasets demonstrate better transcriptome frequency estimation accuracy compared to that of the existing tools which tend to skew the estimated frequency toward super-transcripts.

Monte-Carlo Regression algorithm for isoform frequency estimation from RNA-Seq data

We propose a Monte-Carlo Regression based method for isoform frequency estimation from RNA-Seq reads.

On-line Batch Scheduling in Distributed Optical Networks

Batch scheduling accommodates a group of tasks with the start/end time constraints to maximize the revenue from scheduling tasks over a number of servers, which has been extensively studied in the context of Job-machine scheduling. In optical networks, batch scheduling refers to the process of scheduling a group of data units (i.e., the jobs) that competing for the same set of wavelength channels (i.e., the machines). Classical Job-machine scheduling studies considered both the case of a pure-loss system, and the case with waiting rooms (i.e., buffers), which are generally in the form of Random Access Memory (RAM). In optical networks, the buffering is achieved by feeding the optical signal into a fixed length of fiber, namely Fiber Delay Lines, since optical RAM is not yet available. The unique feature of the discrete and predefined buffering time in fact instantiates a new type of problem, namely Job-machine scheduling with Discrete-time Buffers. In this work, we comprehensively study batch scheduling in optical networks. We show that batch scheduling with and without FDLs corresponds to two different instances of Job-machine scheduling problem. While proving their NP-Completeness, we mathematically model both cases using Integer Linear Programming formulations to provide an optimal scheduling. Given the timeliness request for on-line batch scheduling and the dramatic problem size in optical networks, we also propose polynomial-time heuristic algorithms, which are shown to be near-optimal in our simulations.

Transcriptome assembly and quantification from Ion Torrent RNA-Seq data

We propose novel method for transcriptome reconstruction and quantitation of both known and novel transcripts from Ion Torrent RNA-Seq reads.

Batch scheduling in optical networks

Batch scheduling accommodates a group of tasks with the start/end time constraints to maximize the revenue from scheduling tasks over a number of servers, which has been extensively studied in the context of job-machine scheduling. In optical networks, batch scheduling refers to the process of scheduling a group of data units (i.e., the jobs) competing for the same set of wavelength channels (i.e., the machines). Classical job-machine scheduling studies have considered both the case of a pure-loss system, and the case with waiting rooms (i.e., buffers), which are generally in the form of random access memory (RAM). In optical networks, the buffering is achieved by feeding the optical signal into a fixed length of fiber, namely, a fiber delay line (FDL), since optical RAM is not yet available. The unique feature of the discrete and predefined buffering time in fact instantiates a new type of problem, namely, job-machine scheduling with discrete-time buffers. In this work, we comprehensively study batch scheduling in optical networks. We show that batch scheduling with and without FDLs corresponds to two different instances of the job-machine scheduling problem. While proving their NP-completeness, we mathematically model both cases using integer linear programming formulations to provide an optimal scheduling. Given the timeliness request for on-line batch scheduling and the dramatic problem size in optical networks, we also propose polynomial-time heuristic algorithms, which are shown to be near optimal in our simulations.