Microarray analysis techniques: Difference between revisions

Revision as of 21:13, 1 January 2008

Revision as of 21:28, 1 January 2008

Line 4:

Microarray data analysis involves several distinct steps, as outlined below. Changing any one of the steps has the potential to change the outcome of the analysis, so the MAQC Project<ref>{{cite web | url = http://www.fda.gov/nctr/science/centers/toxicoinformatics/maqc/ | title = MicroArray Quality Control (MAQC) Project | accessdate = 2007-12-26 | author = Dr. Leming Shi, National Center for Toxicological Research | publisher = U.S. Food and Drug Administration }}</ref> was created to identify a set of standard strategies.

==Creating raw data==

Most microarray manufacturers, such as [[Affymetrix]], provide commercial data analysis software with microarray equipment such as plate readers.

Most microarray manufacturers, such as [[Affymetrix]], provide commercial data analysis software with microarray equipment such as plate readers. Raw Affy data contains information about mismatch spots, spots which do not precisely match the target sequence. These can theoretically measure the amount of nonspecific binding for a given target, but some popular approaches like RMA do not take advantage of them.<ref>{{cite journal |author=Bolstad BM, Irizarry RA, Astrand M, Speed TP |title=A comparison of normalization methods for high density oligonucleotide array data based on variance and bias |journal=Bioinformatics |volume=19 |issue=2 |pages=185–93 |year=2003 |pmid=12538238 |doi=}}</ref>. The current Affymetrix MAS5 algorithm, which uses both perfect match and mismatch probes, continues to enjoy popularity and do well in head to head tests.<ref>{{cite journal |author=Lim WK, Wang K, Lefebvre C, Califano A |title=Comparative analysis of microarray normalization procedures: effects on reverse engineering gene networks |journal=Bioinformatics |volume=23 |issue=13 |pages=i282–8 |year=2007 |pmid=17646307 |doi=10.1093/bioinformatics/btm201}}</ref>

==Background correction==

Depending on the type of array, signal related to nonspecific binding of the fluorophore can be subtracted to achieve better results. One approach involves subtracting the average

Line 17 ⟶ 16:

==Aggregation and normalization==

Comparing two different arrays, or two different samples hybridized to the same array generally involves making adjustments for systematic errors introduced by differences in procedures and dye intensity effects. In the case of Affymetrics arrays, there are multiple probesets on an array for the same target sequence, requiring some sort of summarization. The RMA method uses [[median polish]] as opposed to a straight average. Dye normaization for two color arrays is often acheived by [[local regression]]. Quantile normalization, also part of RMA, is one sensible approach to normalize a batch of arrays in order to make further comparisons meaningful. LIMMA provides a set of tools for background correction and scalling, as well an option to average on-slide duplicate spots.<ref>{{cite web |url=http://bioinf.wehi.edu.au/limma/ |title=LIMMA Library: Linear Models for Microarray Data |accessdate=2008-01-01 |format= |work=}}</ref>

MostRaw microarray manufacturers, such as [[Affymetrix]], provide commercialAffy data analysiscontains softwareabout withtwenty microarrayprobes equipmentfor suchthe assame plateRNA readerstarget. RawHalf Affyof datathese contains information aboutare "mismatch spots", spots which do not precisely match the target sequence. These can theoretically measure the amount of nonspecific binding for a given target,. butRMA someis popularan approachesnormalization likeapproach RMAthat dodoes not take advantage of them, but still must summarize the perfect matches through [[median polish]].<ref>{{cite journal |author=Bolstad BM, Irizarry RA, Astrand M, Speed TP |title=A comparison of normalization methods for high density oligonucleotide array data based on variance and bias |journal=Bioinformatics |volume=19 |issue=2 |pages=185–93 |year=2003 |pmid=12538238 |doi=}}</ref>. TheQuantile current Affymetrix MAS5 algorithmnormalization, whichalso usespart bothof perfectRMA, matchis andone mismatchsensible probes, continuesapproach to enjoynormalize popularitya andbatch doof wellarrays in headorder to headmake tests.<ref>{{citefurther journalcomparisons |author=Lim WK, Wang K, Lefebvre C, Califano A |title=Comparative analysis of microarray normalization procedures: effects on reverse engineering gene networks |journal=Bioinformatics |volume=23 |issue=13 |pages=i282–8 |year=2007 |pmid=17646307 |doi=10meaningful.1093/bioinformatics/btm201}}</ref>

The current Affymetrix MAS5 algorithm, which uses both perfect match and mismatch probes, continues to enjoy popularity and do well in head to head tests.<ref>{{cite journal |author=Lim WK, Wang K, Lefebvre C, Califano A |title=Comparative analysis of microarray normalization procedures: effects on reverse engineering gene networks |journal=Bioinformatics |volume=23 |issue=13 |pages=i282–8 |year=2007 |pmid=17646307 |doi=10.1093/bioinformatics/btm201}}</ref>

==Identification of significant differential expression==

Many strategies exist to idenitfy which array probes show an unusual level of over expression or under expression. The simplest one is to call "significant" any probe that differs by an average of at least two fold between treatment groups. More sophicated approaches are often related to [[t test]]s or other mechanisms that take both effect size and variability into account. Curiously, the p-values associated with particular genes do not reproduce well between replicate experiments, and lists generated by straight fold change perform much better.<ref>{{cite journal |author=Shi L, Reid LH, Jones WD, ''et al'' |title=The MicroArray Quality Control (MAQC) project shows inter- and intraplatform reproducibility of gene expression measurements |journal=Nat. Biotechnol. |volume=24 |issue=9 |pages=1151–61 |year=2006 |pmid=16964229 |doi=10.1038/nbt1239}}</ref><ref>{{cite journal |author=Guo L, Lobenhofer EK, Wang C, ''et al'' |title=Rat toxicogenomic study reveals analytical consistency across microarray platforms |journal=Nat. Biotechnol. |volume=24 |issue=9 |pages=1162–9 |year=2006 |pmid=17061323 |doi=}}</ref>This represents an extremely important observation, since the point of performing experiments has to do with predicting general behavior. The MAQC group recommends using a fold change assesment plus a non-stringent p-value cuttoff, further pointing out that changes in the background correction and scaling proess have only a minimal impact on the rank order of fold change differences, but a substantial impact on p values.