SLAS2016 Short Courses
Multiparametric Analysis of Scientific Image Data (Laptop Required; NEW!)
We offer a hands-on course for analyzing multiparametric data typically extracted from images in high content analysis using the open source platform KNIME. The course will show how to transform complex datasets into biological insights using advanced data mining techniques without programming.
Who Should Attend:
- Scientists generating multiparametric data who are struggling to make sense of the mass of parameters they obtain. The course will discuss specifically data emerging from image analysis, but similar concepts can be applied to other rich data sources such as mass spectrometry, transcriptomics or even the analysis of tweets on Twitter.
- The course will discuss advanced aspects of data mining and participants should already have experience in obtaining multi parametric data.
- Participants will need to be equipped with a laptop computer and will receive instructions how to install the required software. No programming skills will be necessary in this course.
How You Will Benefit From This Course:
- Thanks to the open source nature of the software used, participants will be able to directly apply what they have learned in their own laboratory.
- Participants will have an overview of issues and solutions at all stages of the analysis of multiparametric data.
- Participants will join a growing community of data miners using KNIME
- Data inspection and annotation: every data analysis task starts with an inspection of the data including: completeness, type of data, annotation and inspection of biological controls. This section will make participants familiar with the software (data input, data manipulation and first plotting tools).
- Parameter selection: High Content Analysis is complicated because of the plethora of parameters that can be extracted from images. Strategies for parameter choosing will be discussed and applied.
- Clustering, machine learning: multiparametric analysis offers the possibility to sort phenotypes into similar classes indicating possible common mechanisms of either compounds or genes. Several distance measures between n dimensional vectors and algorithms for unsupervised and supervised clustering will be discussed and applied.
- Hit calling: the aim of all screens is to identify active conditions from inactive ones, the course will discuss various ways of identifying hits in multiparametric space.
- Others: depending on interest and time several other common techniques used for plotting or analyzing multiparametric data will be presented such as population analysis, dimension reduction strategies or dose-response curves.
Synergy with other courses:
The course is an ideal extension of the short course "Digital Image Processing and Analysis for the Laboratory Scientist: Theory and Application" by Matthew Fronheiser and Mark Russo. This excellent course teaches how to generate the multiparametric data that is then analyzed in this course. Possible synergies with.
Marc Bickle heads the Technology Development Studio (TDS) of the Max Planck Institute of Molecular Cell Biology and Genetics in Dresden, Germany (http://www.mpi-cbg.de/facilities/profiles/ht-tds.html). The TDS is an open access screening facility specialised in high content screening of RNAi and chemical libraries. The TDS uses and creates open source software for analysing high content data and offers several courses on High Content Screening.
He has been involved in screening both in industry (Aptanomics SA Lyon France) and academia since 15 years and has been developing high content assays since 8 years. His group has been focused on exploiting the richness of multiparametric data and has developed many open source tools for KNIME to carry out sophisticated data mining strategies. Marc Bickle has several years of experience teaching image analysis and data mining with open source software to allow participants to directly apply what they have learned in the course in their own laboratory.