“The greatest value of a picture is when it forces us to notice what we never expected to see.”

– John Tukey

Should be done before any microarray/sequencing experiment

RNA Purity

• Absorbance, optical detection: 260:280 ratios and 260:270 ratios
• Pure RNA has 260:280 ratio is ~2.0
• Contaminations (protein, DNA, phenol) will affect absorbance ratios

• Capillary gel electrophoresis: 28S:18S ribosomal RNA ratio
• Pure RNA has 28S:18S ratio >2

• Capillary gel electrophoresis: 28S:18S ribosomal RNA ratio
• Pure RNA has 28S:18S ratio >2

• Spot/Probe level: poor quality of gene expression measurement on one particular array;

• Gene level: poor quality of the expression measurement for a single gene across all arrays;

• Array level: poor quality of all spots on a particular glass slide or GeneChip.

• Number of spots excluded due to quality problems
• Number of saturated pixels
• Amount of normalization required
• Present/Absent calls for ribosomal RNAs
• 3′:5′ Ratios
• 2D Spatial images
• Boxplots
• MA Plots
• Linearity of signal intensities

Problematic spots should be flagged and omitted from subsequent analyses

• Spots with too many saturated pixels
• No hybridization
• Negative signal after background adjustment

• MA plots used to compare two microarrays (two vectors of intensities)
• MA plots show the dependence between average log2 intensity (x-axis) and ratio of the two intensities (y-axis)
• Better use of the plot's real estate - outliers in logarithm scale spreads the data from the lower left corner to a more centered distribution in which the prosperities of the data are easy to analyze.
• Easier to describe the fold regulation of genes using a log scale. In log2 space, the data points are symmetric about 0.