Neural Designer Using machine learning methods lets you discover intricate relationships and recognize complex patterns from microarray data.

A microarray is an experimental format based on the synthesis or attachment of probes on a solid substrate intended to be exposed to the target molecules.

The fluorescence indicates the level of hybridization between the two, and it shows the level of expression of the probe about the target molecule.

Different probes can be analyzed using microarrays. Some examples are DNAs, RNAs, proteins, tissues, peptides…

These are some of the microarray applications:

• Study of genes under different conditions. One example would be healthy versus ill.
• We are identifying genes characterizing a disease.
• Predict the response of a certain treatment.
• Environmental stress responses.

One of the most famous applications of machine learning is the molecular classification of cancer by gene expression conducted by Golub et al.

A microarray data set stores experimental data in a matrix, where rows usually correspond to gene names or probe identifiers and columns to sample identifiers.

The scale and complexity of microarray data sets is increasing exponentially. Indeed, large-scale experiments involve monitoring the expression levels of thousands of probes simultaneously.

All of the information about a microarray experiment must be recorded systematically
to generate meaningful data sets.

Fortunately, there are also a lot of public databases for arrays, like Array Express at the European Bioinformatics Institute or the NCBI´s Gene Expression Omnibus GEO.

There has also been an effort to standardize the microarray data sets so the researchers can share common knowledge and compare experiments. The Microarray Gene Expression Data Society is leading this effort.

These are some of the benefits that machine learning can provide in microarrays:

ANALYZE GENES

Analyze the changes in the genes to find patterns and determine if they are in a standard or diseased state.

DIFFERENTIATE GENE´S STAGES

Determine the conditions that make a gene change from normal to disease state.

PREDICT THE FUTURE STAGES

Develop a model that can recognize the changes in the genes and predict their state.

PREVENT DISEASES

Use the predictive model for preventive medicine and early diagnosis.

Machine learning can help us discover relationships between genes and diseases and between genes from large data sets.

Neural Designer is a machine learning software that implements neural networks, a technique capable of analyzing large amounts of data and finding complex relationships between them.

The tool provides you with different methods to analyze relations between the state of the genes and the environmental conditions, model selection algorithms for the best design of the neural network or the most important variables, and a set of testing methods to check the correct functioning of the predictive model.

You can see how Neural Designer can solve this example and find correlations between the leukemia class and different genes.

Thanks to the input selection algorithms, Neural Designer can find the most influential genes in leukemia.

Conclusions

The study of microarrays and gene expression can bring useful knowledge to very complex subjects, and it is a tool healthcare researchers are using to learn about diseases and how to treat them.

Relevant documentation:

If you want to learn how to use Neural Designer, you can do that by reading the user´s guide or practicing with the examples.

You can also learn more about neural networks by reading this guide.

References:

• Golub, T. R., Slonim, D. K., Tamayo, P., Huard, C., Gaasenbeek, M., Mesirov, J. P., … & Bloomfield, C. D. (1999). Molecular classification of cancer: class discovery and class prediction by gene expression monitoring. science, 286(5439), 531-537.
• Kuo, W. P., Kim, E. Y., Trimarchi, J., Jenssen, T. K., Vinterbo, S. A., & Ohno-Machado, L. (2004). A primer on gene expression and microarrays for machine learning researchers. Journal of Biomedical Informatics, 37(4), 293-303.
• Cho, S. B., & Won, H. H. (2003, January). Machine learning in DNA microarray analysis for cancer classification. In Proceedings of the First Asia-Pacific bioinformatics conference on Bioinformatics 2003-Volume 19 (pp. 189-198). Australian Computer Society, Inc.