Using Big Data to Predict Toxicity of Chemicals Can Save Animals

There is a large amount of available data that can be used to predict chemical toxicity without carrying out animal tests, thanks to international data-sharing projects. The enormous volume of these databases makes it difficult to use conventional data-analysis tools when processing them. Recent advances in big-data analytics, however, provide new methods for predictions of chemical toxicity.
Recently, experts gathered at the Indian Institute of Technology Delhi for a national event, “Breaking Barriers through Bioinformatics and Computational Biology.” They shared information on the latest development in computational biology.
PETA India described how companies could use big data to reduce animal testing. It also asked private and government organizations to use big-data analytics methods. A PETA India poster discussed the disadvantages of traditional animal-based methods to determining the toxicity of chemicals, the progress on data-sharing, and how data in public repositories can be used to make models that can predict the toxicity level of chemical compounds. The steps that regulatory and government authorities can take to adopt big data and reduce animal testing were also discussed.
According to PETA India’s Dr. Rohit Bhatia, using big advanced data analytics approach to predict toxicity save money, time, and lives of many animals.

BioUML: An Open Source Software Platform for System Biology


BioUML is an open source and extensible software framework for data analysis from advanced computational biology developed by researchers from the Institute of Systems Biology. The platform is available online and is used in research labs for the discovery of disease origins and prevention. The platform aims at covering all areas of computational applications in systems biology and bioinformatics.
Currently, BioUML include three versions: BioUML Server, BioUML Workbench, and BioUML Web Edition. BioUML Server provides access to data and analysis techniques. BioUML Workbench is a Java application that works standalone for the platform server edition. Lastly, BioUML Web Edition is a web browser that offers most of the BioUML workbench functionality.
Since 2003, the platform has been developed constantly and provides data analysis and visualizations for researchers involved in molecular biology research. It allows scientists to comprehensively describe biological systems function and structure including tools needed to make findings related to metabolomics, transcriptomics, proteomics, and genomics.
High-throughput sequencing and other next-generation sequencing methods create big data. BioUML platform disseminates, study, and produce simulations and visualizations, facilitates parameter fitting and supports numerous other analysis techniques needed to deal with large amounts of data.

Scientists Identify 760 Genes that are Key to Cancer Growth


In one of the biggest efforts to create a detailed catalog of genetic susceptibilities in cancer, scientists from Harvard and Dana-Farber Cancer Institute and the Broad Institute of MIT have identified over 760 genes upon which many types of cancer are dependent for their survival and growth.
According to the researchers, many of these “dependencies” are specific to particular cancer types. About 10% of them, however, are common across numerous cancers. This suggests that relatively few therapies targeting these main dependencies may each hold promise for fighting several tumors.
To get these findings, the scientists carried out genome-wide RNA interference screens on 501 cell lines that represented over 20 types of cancer, silencing over 170000 genes separately in each line to find genetic dependencies that are unique to cancerous cells.
Cancer cells can have various genetic errors, ranging from small mutations to extensive exchanges of DNA between chromosomes. When an error affects a critical gene, a tumorous cell compensates by changing the activity of other genes, frequently creating a dependence on such changes to persist.
Finding these dependencies does not only offer an opportunity for the researcher to gain a better insight into cancer biology but also enable them to determine new therapeutic targets.
First draft of genome-wide cancer ‘dependency map’ released

Using Google Searches to Track Infectious Diseases

New research shows Google searches can help health officials track infectious diseases. A team from Harvard University has quickly and accurately tracked dengue fever in some undeveloped nations by using a mathematical model that combines clinical government data and Google searches. The research was published in the journal “PLOS Computational Biology” last week.
Dengue fever is a mosquito-borne disease that has flu-like symptoms such as nausea, muscle aches, and headaches. It is estimated that there are 390 million dengue infections per year. More than 2.5 billion people (30 % of the world’s population) are at risk of this infection.
Dengue mostly occurs in less developed countries. Therefore, tracking this disease is difficult because these countries do not have reliable and effective surveillance systems for it. In these regions, governments rely on hospital reports which are often delayed and have many regions.
The new research built on a previous method to track the flu in the U.S in 2015. The researchers believe that the new study restores hope that online searches can help track infectious diseases after earlier efforts such as Google Dengue Trends and Google Flu Trends returned mixed findings and were discontinued.
The researchers modified their mathematical modeling tool and tracked dengue activity in Taiwan, Singapore, Thailand, Brazil, and Mexico. Google’s “Trends” tool was used to track searches made by individuals in these countries. The researchers found their method produced more precise estimations than other techniques used in other regions.

What is Computational Pharmacology?

Computational methods based on mathematics and statistics are permeating all areas of pharmacology. Theoretical and computational methods are revolutionizing drug discovery and pharmacology. In the drug discovery, predicting, modeling, and simulating therapeutic agents and their relations with target molecules is a great step.
From a computational biology perspective, computational pharmacology is the study of genomic data effects to find relations between diseases and specific genotypes and then screening drug data. The pharma industries need a shift in methods to analyze drug data. In the past, pharmacologists have been using Microsoft Excel to compare genomic data and chemical associated with drug effectiveness. However, the industry is now at what is known as the Excel barricade, arising from the limited cells accessible on a spreadsheet. As a result, scientists have developed computational techniques to analyze massive data sets.
Analysts predict that if medications fail due to intellectual property, computational biology will be needed to replace existing drugs on the market. Currently, many pharmaceutical companies need more competent analysts of the massive data sets who can help in the production of new drugs. Therefore, students in computational biology should be encouraged to pursue careers in computational pharmacology.

Network of the Brain May Allow Researchers to Understand Neurological Disorders Better

According to a recent study by Weill Cornell Medicine and the University of California, San Francisco, a deeper understanding of neurons connectivity network of the brain could enable scientists to predict spatial patterns of the brain and identify processes that relate to neurological disorders.
In their study that was published in PLoS Computational Biology on 22th June, the researchers used mathematics and Diffusion-Tensor MRI ( a form of magnetic resonance imaging) to better understand how the connections between the brain’s network of fibers and its deep white matter are affected by neurological disorders. They discovered sub-networks that make the connectome, eigenmodes, whose role is to communicate information from one part of the brain to another.
According to senior author Ashish Raj, an associate professor of neuroscience and computer science in radiology in the Feil Family Brain and Mind Research Institute, once an individual understands the eigenmodes, he can start understanding the processes that occur in the brain and the patterns of activity to expect.
Raj described the connectome of the brain as series of tubes through which information runs. The connectome’s components parts (eigenmodes) are “similar to guitar’s vibrations, “he said. The researchers carried out MRIs on ten healthy participants and discovered that eigenmodes are pervasive with high overlap between participants and between scans of the similar participants carried out on different days.

Computational Neuroscience


Also known as theoretical neuroscience, computational neuroscience studies functions of the brain in terms of the information processing features of the structures that are part of the nervous system. Known as an interdisciplinary of computational science, it links the various fields of neuroscience, psychology, cognitive neuroscience with physics, mathematics, computer science and engineering.

Computational neuroscience is not only different from psychological connectionism but also learning theories of disciplines such as computational learning, neural networks, and machine learning. This is because it focuses mainly on descriptions of biologically and functional realistic neurons and their dynamics and physiology.

Computational neuroscience uses theoretical tools to predict, interpret, or explain experimental data and the mechanisms that underlie it. In neuroscience, some examples of the use of quantitative methods include:

Advanced data analysis
Creating tools that help to get more information and understanding from neural data on many scales.

Dynamical and statistical modeling
Use of data to develop models whose functions is to look at the role that is played by specific neural mechanisms.

Create novel models that bring together or explain disparate observations.
Identifying minimal dynamics or structure that explain an important neural phenomenon.
Identifying algorithmic/ computational motifs and principles that explain and optimize neural function.



Researchers Develop a New 3-D Model That Predicts Best Planting Practices

As farmers plan to plant, various questions come to mind: How does changing row spacing affect yields? How many plants germinated in my field last year? Is there any difference if I plant my rows east to west or north to south? Now a new3-D model can answer those by relating numerous virtual fields with diverse planting densities, orientations, and row spacing.
The Partner Institute for Computational Biology in Shanghai and the University of Illinois developed a computer model that calculates the yield of different crop cultivars in various planting conditions. The model shows the growth of 3D plants, integrating models of the biophysical and biochemical processes that underlie productivity.
Working closely with the University of Sao Paulo in Brazil, researchers from both universities used the model to answer questions for sugarcane producers. Double row spacing is recommended for sugarcane plants. However, this type of spacing sacrifices up to 10 percent of yield. Using the model, the researchers discovered that double-row spacing costs farmers 10% of yield compared to traditional row spacing.
Researchers predict that this model could be applied to various crops to predict best planting strategies for specific environments.

Understanding Evolutionary Computation

Evolutionary Computation is a collective term for several problem-solving methods based on biological evolution principles, such as genetic inheritance and natural selection. In computer science, evolutionary computation is a family of algorithms inspired by biological evolution and artificial intelligence studying these algorithms.
In evolutionary computations, the first set of candidate solutions is produced and iteratively updated. A new generation is generated by stochastically eliminating less needed solutions, and introducing some small random changes. Biologically, a population of solutions is exposed to artificial selection (or natural section) and mutation. Consequently, the population gradually evolves to increase in fitness.
Evolutionary computing methods mainly involve metaheuristic optimization algorithms. The field includes ant colony optimization, artificial bee colony algorithm, artificial immune systems, artificial life, bees algorithm, cultural algorithms, differential evolution, dual-phase evolution, evolutionary algorithms, evolutionary programming, evolution strategy, gene expression programming, genetic algorithm, genetic programming, grammatical evolution, harmony search, learnable evolution model, learning classifier systems, particle swarm optimization, self-organization swarm and intelligence.
Evolutionary computation methods can generate highly optimized solutions in many problem settings. Numerous variants and extensions exist, suitable for more specific problems and data structures. Sometimes, evolutionary computation is also used in evolutionary biology. For instance, it is used in the silicon experimental process to study common features of general evolutionary processes.

A New Software Tool that Could Help Medics Diagnose Genetic Diseases

Many diseases that are caused by genetic mutations are challenging to diagnose. Currently, sequencing of the entire genome of a patient—the genome part used to build proteins—offers a strategy to identify culprit mutations and consequently make an accurate diagnosis. Unfortunately, the software that is needed to examine these sequences is often too complex or costly for many medics to use.
To solve these problems, Cardenas’ team made a new software tool that is known as Mendel, MD. The scientist developed this tool specifically for easy use by doctors, free of charge. Using web-based interface, users upload the genome sequence of a patient, and the sequence is examined and filtered using different computational database and tools of disease-causing mutations. As a result, users get a list of candidate mutations that is clinically examined to arrive at a final diagnosis.
The scientists validated the new tool using previously-published clinical cases. The researchers also had it tested by students at their university, as well as at the Children’s University Hospital in Dublin, and Núcleo de Genética Médica (GENE) in Brazil. The results show that the tool is simple, reliable and efficient in detecting disease-causing mutations in patients.