Implementation of TaqMan SPPs
From LipidomicsWiki
After the definition of the SPP´s, Integromics SL has developed and implemented algorithms, written in R code, which will allow experimental scientists to obtain reliable results from RT-PCR raw data. The final result is a package which makes use of those algorithms and statistical libraries from R environment. The implemented algorithms are described bellow:
Contents |
Data importing:
Raw data generated with TLDA cards (from Applied Biosystems) needs to be imported as the input format defined in the SPP’s, as well as experimental design information. Data will be transformed to a large format matrix and the number of plates used in the study will be automatically recognized.
After loading files, defining the experimental design and renaming samples, the data will be represented by a gene by samples matrix in which each entry ai.j of the matrix is the Ct value of gene i in the sample j.
Quality control of Biological Replicates:
Two types of plots will permit the user to visualize Ct values distribution among biological groups and samples. A third visual representation will be a colored heat map which is obtained from a correlation matrix (using Pearson distance measurement).
Detectability criteria:
Ct values above a defined threshold may be unreliable. The software includes an algorithm which flags transcripts as “not detected” and returns this information on the output table of the relative quantification experiment.
Normalization:
In order to estimate the best normalization genes, Integromics software offers one method to evaluate gene expression stability called GeNorm (Vandesompele J. et al., 2002). Such a method estimates the pairwise variation (standard deviation of the logarithmically transformed expression ratio) of a control gene with all the other control genes of the experiment. From this, a gene stability measure M is calculated as the average pairwise variation. Genes with the lowest M values have the most stable expression.
Therefore genes for normalization could either be selected by inspecting the profiles in a plot or by performing GeNorm test.
Furthermore, the algorithm is able to suggest a computed reference gene whenever more than one gene are selected for normalization.
At this stage, the data will be represented by a gene by samples matrix in which each entry ai.j of the matrix is the calculated ΔCt value of gene i in the sample j.
Relative quantification:
The implemented algorithm use linear models for analyzing designed experiments and the assessment of differential expression. A t-test is then performed in order to get the statistical significance of the results, including a method (Benjamini-Hochberg) to adjust for false discovery rate.
Clustering:
This algorithm generates a distance matrix (using Euclidean distance measurement) and clusters the data using the complete linkage method.
The complete linkage method links pairs of objects that are close together into binary clusters (clusters made up of two objects). Distance between clusters will be defined as the distance between the most distant pair of objects, one from each cluster. At each stage, the clusters for which such a distance is the minimum will be merged.
So in this way, newly formed clusters will be linked to each other to create bigger clusters. Once all the objects in the original data set are linked together, a heat map will be returned.
Processing Pipeline
Processing Pipeline for TaqMan SPPs
