labrlfree非标记定量蛋白质组学的分析方法.pptx

Click to edit Master title style,Click to edit Master text styles,Second level,Third level,Fourth level,Fifth level,6/8/2014,#,Introduction,Label Free Quantitation in MaxQuant,GlyGly Diagnostic,Peaks,New Annotation Visualization Options,Label Free Quantitation(LFQ),Compare quantitative measures of proteins/peptides BETWEEN mass spec runs to identify significant changes,Comparison is WITHIN the mass spec run for,SILAC,iTRAQ,and IPTL,No chemical or metabolic labeling required,Compatible with virtually any sample,Spectral Counting,#MS/MS identified for each protein,Very crude,Easily implemented,Somewhat,sample dependent,Spectral counts are sometimes normalized across,mass spec runs,Extracted Ion Chromatogram(XIC),Sum of areas under the curves of MS1,chromatograms of each peptide identified,Requires,advanced processing software,Somewhat independent of the sample,Label Free Quantitation in MaxQuant,SILAC ratio calculation requires the comparison of,XICs between the label states in a mass spec run,Can use same processing to calculate XICs of peptides in each run and then compare across mass spec runs,Need to map from one mass spec run to others,Easy if MS2 identified in both runs,Harder,if not identified in both runs,MaxQuant has option to match identified peaks to unidentified,peak using retention time and accurate mass,LFQ Options in MaxQuant,Peptide Level,Raw intensity,only option,Protein Level,Raw intensity,MS2,counts,LFQ value,No publication yet,Calculate global raw file normalization factor,iBAQ,Normalize intensities by number of observable peptides,Benchmarking,LFQ Options,Triplicate,yeast analyses,Pearson Correlation,Typical correlation measure,Sensitive to outliers,Spearman,Rank Correlation,Only looks at ranking,Insensitive,to outliers,Ratio 95%Confidence,Interval,Tells us the typical range,of ratios we would expect to see for values that are actually unchanged,Peptide LFQ Performance,Protein Raw,Intensity LFQ Performance,Protein Spectral Counts LFQ Performance,Protein iBAQ LFQ,Performance,Protein LFQ Performance,LFQ,Summary,Now possible to get reasonable estimates of relative abundances between,samples at both the peptide and protein level,Protein 95%CI at,1.5 fold change for 1:1,Peptide 95%CI at 3 fold change for 1:1,Might consider using these techniques to measure large changes in modification status,For,example,protein footprinting,Ubiquitin Modification Data Mining,Ubiquitin modifications,result in a diGlycine modification on Lysine,Geoff had produced a large dataset potentially containing many ubiquitin modifications,Unfortunately,the error rate for these modifications,was quite high,Wondered if there were characteristics of modified peptides that made them difficult to identify in a standard database search,Analyzed peptides with modified Lysines at various positions,Sonja identified peaks that would correspond to backbone fragmentation between K-G(and between G-G),Ubi,Modifications,The K-G and G-G bonds are both peptide bonds and are therefore susceptible to fragmentation,We should see diagnostic fragment ions g1 and g2 that correspond to the peptide without one or two Glycines,G,G,A-A-K-A-T-R,g2,g1,Ubi Modified Spectra,g2,g1,Ubi Modified Spectra Stats,What are the typical charge states of the fragment ions missing Glycines?,How do these charge states depend on the precursor charge state?,How intense,are the diagnostic fragment ions?,Are there any explanation for modified peptides missing diagnostic peaks?,Use Tsui-Fen,and Gygi ubiquitin-enriched data sets to mine the data for answers,Ubi Modified Spectra,Stats,In general,modified spectra tend to have a higher charge state,than unmodified spectra,Ubi Diagnostic Ion Statistics,Tsui-Fen Dataset,Ubi Diagnostic Ion Statistics,Tsui-Fen Dataset,Ubi Diagnostic Ion Statistics,Tsui-Fen Dataset,Ubi Diagnostic Ion Statistics,Tsui-Fen Dataset,Ubi Diagnostic Ion Statistics,Tsui-Fen Dataset,Gygi Dataset,Ubi Summary,There does appear to be a diagnostic ion in many modified peptides,How does the charge state of the diagnostic ion correlate with the precursor ion?,Is there any correlation between the peptides that do not have an observable(or strong)diagnostic ion,Unclear how specific mass spec settings may affect the formation of the diagnostic ions,Visualizing DAVID Analysis,In a recent Mann paper,they used a non-parametric statistical test to determine if the rank of the protein ratios with a particular GO annotation were significantly higher or lower than the other proteins,Mann-Whitney-Wilcoxon Test,No relation to Matthias Mann,Additionally they used a“violin plot”to display the distribution of the ratios,Developed scripts to perform similar calculations on data for Bobby,Volcano Plot of DAVID Analysis,Violin Plot of Significant Terms,SILAC Analysis Summary,New plots help to detect interesting terms where there are few individual proteins that are significantly different,but the majority of the proteins are slightly different,May not work for terms where half of the proteins are up and half are down,Violin plots help to display the distribution of protein ratios within an annotation term,A histogram of the ratios would also be appropriate,