Oligonucleotide microarrays for quantifying mRNA abundance Statistics 246 Spring 2006 Week 8 Lecture 2 1 Gene expression the central dogma DNA CCTGAGCCAACTATTGATGAA transcription mRNA CCUGAGCCAACUAUUGAUGAA translation Protein PEPTIDE 2 Our goal here To measure gene expression as indicated by mRNA abundance on a genome wide transcriptome wide scale Most people would prefer to measure gene expression via protein abundance but that is currently impossible on a genome wide proteome wide scale However attempts at doing so are moving along rapidly 3 Recall DNA structure C G 3 H bonds A T 2 H bonds 4 Hybridization This is a technique which exploits that potent feature of the DNA duplex the sequence complementarity of the two strands When DNA is heated it melts or denatures i e the H bonds break and the two strands come apart Remarkably when cooled the single stranded DNA can then reassemble renature anneal hybridize with perfect fidelity from the separated strands into double stranded DNA The melting temperature of a DNA molecule is denoted by Tm This depends primarily on G C content of the sequence but more generally on the entire sequence of the molecule salt concentration pH and other aspects of 5 the solution 6 7 8 DNA probes We now introduce the DNA probes used in the hybridization reaction to generate the data we use to quantify mRNA aboundance in our samples Here probe is a general term for single stranded s s DNA sequences that are complementary in the Watson Crick sense A T G C to their s s target sequences In a hybridization reaction complementary s s sequences find one another and base pair that is form a double stranded d s hybrid We say one sequence binds to the other In the reaction more perfect matches generally leads to stronger binding Also mismatches near the center generally reduce binding more than mismatches near ends 9 More on probe binding and melting temp TAGCCATCGGTAAGTACTCAATGAT Perfect Match pairing binds best giving ATCGGTAGCCATTCATGAGTTACTA highest melting temperature Tm the more G Cs the higher the Tm TAGCCATCGGTATGTACTCAATGAT Mismatch probe at one base in the ATCGGTAGCCATTCATGAGTTACTA center generally has most effect on binding among single mismatches TAGCCATCGGTAAGTACTCAATGAT Mismatch at one base at one end CTCGGTAGCCATTCATGAGTTACTA generally has least effect on binding among single mismatches TAGCCATCGGTAAATACTCGATGAT Many mismatches leads to very poor ATAGGTCGCCATTCATCAGTTAATA binding low Tm 10 The Affymetrix chip 3 probe selection Probes are 25 mers selected from the complement of a target mRNA sequence Frequently probes overlap 5 50K target fragments are interrogated together bin probe sets of 11 20 probes chosen from the 3 end of transcripts Affymetrix typically uses both PM and MM probes for why see later Recently arrays with probes in all exons and elsewhere have 11 become available and these are PM only Probe selection rules When there is a choice to be between possible probes the selection will me made according to a variety of criteria weighted and combined by the chip designer One will be the uniqueness of the probe sequence in the relevant genome how close are the the closest alternatives Another will be the Tm which reflects binding affinity of probe for target A third might be be the uniformity of the Tm s across the entire set of probes As we will see the chip is hybridized at one temperature so all probes must perform at that temperature A fourth might be the tendency of the probe to form secondary structures to be avoided Spacing is also important overlap is not desirable if it can be avoided and so on In general having probes near the 3 end increases the chance of their being unique to that gene 12 The Affymetrix SNP chip cont Here the probes are fixed to the chip During the hybridization reaction fluorescently labelled target DNA see the s fragments find their WatsonCrick complements and base pair with them After the hybridization reaction is completed unbound target DNA is washed off Then the chip is stained and scanned and a fluorescent readout for every feature on it gives us our data A larger read out means more d s DNA bound to the feature i e more of that probe sequence in the target DNA The above13and most of the figures which follow are from the Affymetrix web site Affymetrix synthesizes probes onto the chip 14 Oligonucleotide arrays an overview GeneChip Probe Array Hybridized Probe Cell Single stranded labeled RNA target Oligonucleotide probe 1 28cm 18 18 m or smaller 106 107 copies of a specific oligonucleotide probe per feature 450 000 different probes Image of Hybridized Probe Array 15 GeneChip eukaryotic expression assay 16 17 18 Processing step RNA isolation QA QC procedures Gel electrophoresis OD cDNA synthesis Gel electrophoresis Biotin labeled cRNA synthesis Gel electrophoresis OD cRNA fragmentation Gel electrophoresis Hybridization to array Array wash and stain Array scanning Examination of the intensity of image Image analysis Examination of chip quality indicators 19 and control probe sets Obtaining the data RNA samples are prepared labeled hybridized with arrays and stained as just described Arrays are scanned and the resulting image analyzed we omit the details Approximately 30 50 pixels per probe cell now fewer are summarized by their 75th tile after removal of outer perimeter pixels This is the probe cell s intensity Of interest is to find a way to combine probe cell intensities for a given gene to produce an index of expression an indicator of abundance of the corresponding target mRNA We discuss this later 20 Cartoon version of the process courtesy of Rafael Irizarry followed by some real images 21 After fragmenting before target labelling Target fragments Probe sequences fixed to chip 22 After labelling before hybridization Labelled s s target mRNA actually cRNA 23 After hybridization and wash 24 Ideal quantification 4 2 0 3 25 Actual quantification Imaging by a scanner 26 Chip dat file checkered board oligo B2 27 Chip dat file checkered board close up 28 Chip dat file checkered board close up w grid 29 More visually R hi B lo PM MM Probe2 Probe1 5 30 3 Enlargement of the preceding Features PM or MM are pixellated A probe cell summary value is typically the 75th ile after a boundary of pixels is removed 31 The first low level statistical problem Each gene is represented by 11 20 pairs PM and MM of probe intensities Each array has 8K 40K genes Usually we have multiple arrays We need a measure for each
View Full Document
Unlocking...