Deep learning for computational biology

Angermueller, Christof; Pärnamaa, Tanel; Parts, Leopold

doi:10.15252/msb.20156651

Cited by 1,260 publications

(934 citation statements)

References 94 publications

Supporting

Mentioning

921

Contrasting

Unclassified

Order By: Relevance

“…For example, the newest computer hardware technology, such as graphics processing units (GPUs) and multicore central processing units (CPUs), can be implemented to parallelize calculations (Ayres et al 2011). In addition, new techniques from Artificial Intelligence and deep learning could be applied to 'text mine' genomes which may yield cases of hybridization where we had not expected them (Fogel 2008;Angermueller et al 2016;Leung et al 2016). …”

Section: Sphyrapicus Variusmentioning

confidence: 99%

Avian introgression in the genomic era

et al. 2017

View full text Add to dashboard Cite

Introgression, the incorporation of genetic material from one (sub)species into the gene pool of another by means of hybridization and backcrossing, is a common phenomenon in birds and can provide important insights into the speciation process. In the last decade, the toolkit for studying introgression has expanded together with the development of molecular markers. In this review, we explore how genomic data, the most recent step in this methodological progress, impacts different aspects in the study of avian introgression. First, the detection of hybrids and backcrosses has improved dramatically. The most widely used software package is STRUCTURE. Phylogenetic discordance (i.e. different loci resulting in discordant gene trees) is another means for the detection of introgression, although it should be regarded as a starting point for further analyses, not as a definitive proof of introgression. Specifically, disentangling introgression from other biological processes, such as incomplete lineage sorting, remains a challenging endeavour, although new techniques, such as the D-statistic, are being developed. In addition, phylogenetics might require a shift from trees to networks. Second, the study of hybrid zones by means of geographical or genomic cline analysis has led to important insights into the complex interplay between hybridization and speciation. However, because each hybrid zone study is just a single snapshot of a complex and continuously changing interaction, hybrid zones should be studied across different temporal and/or spatial scales. A third powerful tool is the genome scan. The debate on which evolutionary processes underlie the genomic landscape is still ongoing, as is the question whether loci involved in reproductive isolation cluster together in 'islands of speciation' or whether they are scattered throughout the genome. Exploring genomic landscapes across the avian tree of life will be an exciting field for further research. Finally, the findings from these different methods should be incorporated into specific speciation scenarios, which can consequently be tested using a modelling approach. All in all, this genomic perspective on avian hybridization and speciation will further our understanding in evolution in general.

show abstract

Section: Sphyrapicus Variusmentioning

confidence: 99%

Avian introgression in the genomic era

et al. 2017

View full text Add to dashboard Cite

show abstract

“…Life Sciences have long been one of the key drivers behind progress in AI, and the vastly increasing volume and complexity of data in biology is one of the drivers in Data Science as well. Life Sciences are also a prime application area for novel machine learning methods [2,51]. Similarly, Semantic Web technologies such as knowledge graphs and ontologies are widely applied to represent, interpret and integrate data [12,32,61].…”

Section: Data and Knowledge In Research -The Case Of The Life Sciencesmentioning

confidence: 99%

Data Science and symbolic AI: Synergies, challenges and opportunities

Hoehndorf

Queralt-Rosiñach

2017

View full text Add to dashboard Cite

Editor: Tobias Kuhn (https://orcid.org/0000-0002-1267-0234) Solicited reviews: anonymous reviewer; Rinke Hoekstra (https://orcid.org/0000-0001-7076-9083); Agnieszka Ławrynowicz (https://orcid.org/0000-0002-2442-345X); Honghan Wu (https://orcid.org/0000-0002-0213-5668) Received 21 February 2017 Accepted 27 April 2017Abstract. Symbolic approaches to Artificial Intelligence (AI) represent things within a domain of knowledge through physical symbols, combine symbols into symbol expressions, and manipulate symbols and symbol expressions through inference processes. While a large part of Data Science relies on statistics and applies statistical approaches to AI, there is an increasing potential for successfully applying symbolic approaches as well. Symbolic representations and symbolic inference are close to human cognitive representations and therefore comprehensible and interpretable; they are widely used to represent data and metadata, and their specific semantic content must be taken into account for analysis of such information; and human communication largely relies on symbols, making symbolic representations a crucial part in the analysis of natural language. Here we discuss the role symbolic representations and inference can play in Data Science, highlight the research challenges from the perspective of the data scientist, and argue that symbolic methods should become a crucial component of the data scientists' toolbox.

show abstract

“…As matter of fact, current researches mainly focus on the theory framework and software realization of deep learning neural network [13], [14], while the method for hardware implementation is still lacking. Furthermore, under the constraint of the traditional computing methods.…”

Section: Hardware Comparison and Application Forecastmentioning

confidence: 99%

Low Cost and Fault Tolerant Parallel Computing Using Stochastic Two-Dimensional Finite State Machine

Wang

Yuan

Chen

et al. 2017

IEICE Trans. Inf. & Syst.

View full text Add to dashboard Cite

SUMMARYHardware implementation of neural networks usually have high computational complexity that increase exponentially with the size of a circuit, leading to more uncertain and unreliable circuit performance. This letter presents a novel Radial Basis Function (RBF) neural network based on parallel fault tolerant stochastic computing, in which number is converted from deterministic domain to probabilistic domain. The Gaussian RBF for middle layer neuron is implemented using stochastic structure that reduce the hardware resources significantly. Our experimental results from two pattern recognition tests (the Thomas gestures and the MIT faces) show that the stochastic design is capable to maintain equivalent performance when the stream length set to 10 Kbits. The stochastic hidden neuron uses only 1.2% hardware resource compared with the CORDIC algorithm. Furthermore, the proposed algorithm is very flexible in design tradeoff between computing accuracy, power consumption and chip area.

show abstract

Deep learning for computational biology

Cited by 1,260 publications

References 94 publications

Avian introgression in the genomic era

Avian introgression in the genomic era

Data Science and symbolic AI: Synergies, challenges and opportunities

Low Cost and Fault Tolerant Parallel Computing Using Stochastic Two-Dimensional Finite State Machine

Contact Info

Product

Resources

About