Transcription
1-M Gerstein&P EmaniLectures.GersteinLab.orgComparing RNA& Protein Abundance
uORFs (Current result) Affect translation &relationship betweenprotein & RNA Feature integration to findsmall subset of uORFsthat most alter translation Future Direction:Protein v RNA using matchedsamples in the Brainspandataset single-cell data2- Past Context:to work in the Center Quantifying the moderatestatistical correlationbetween protein & RNA PARE server EMpire (Current result) Leveraging the correlation tobetter assign peptides toisoforms EM algorithm better assignsdominant isoforms, withgreater interpretabilityLectures.GersteinLab.orgOutline: Comparing Protein & RNA Abundance
Why relate amounts of protein & mRNA?Gene expression major place for regulation(easy to measure)vs.Concentration of protein major determinant of activityAt steady state: Pi kd,iwhere ks,i and kd,i are the protein synthesisand degradation rate constants[Greenbaum et al. Bioinformatics 2002, 18, 587]Outliers from trend interesting3-dPi ks,i [mRNAi ] – kd,i Pidtks;i [mRNAi ]Lectures.GersteinLab.orgExpectations from simple kinetic models:
[Greenbaum et al. Bioinformatics 2002, 18, 587]4-r 0.67Lectures.GersteinLab.orgEarly result on mRNA vs Protein, using 2D gels
PAREOpen-source codeDownloadableAnalyze all or GOsubsetLog-log plot ofcorrelation-linear fit-outliers labeled5-[Yu et al., BMC Bioinfo. '07]Lectures.GersteinLab.orgCalculation ofmutual information
uORFs (Current result) Affect translation &relationship betweenprotein & RNA Feature integration to findsmall subset of uORFsthat most alter translation Future Direction:Protein v RNA using matchedsamples in the Brainspandataset single-cell data6- Past Context:to work in the Center Quantifying the moderatestatistical correlationbetween protein & RNA PARE server EMpire (Current result) Leveraging the correlation tobetter assign peptides toisoforms EM algorithm better assignsdominant isoforms, withgreater interpretabilityLectures.GersteinLab.orgOutline: Comparing Protein & RNA Abundance
[Carlyle, Kitchen et al. (2018) Journal of Proteome Research]7-Lectures.GersteinLab.orgIntegration of RNA-seq and Proteomic Data for Isoform Interpretation
Challenge for Isoform-Level Interpretation of Proteomics DataMultimapping Different assays reflecting expression at various levels More reads at earlier stage assay (RNA-Seq FP MS)[Carlyle, Kitchen et al. (2018) Journal of Proteome Research]8-Lectures.GersteinLab.org Leverage other assays for better estimation
[Carlyle, Kitchen et al. (2018) Journal of Proteome Research]9-Lectures.GersteinLab.orgEMpire (Expectation Maximisation Propagation of Isoform abundance from RNA Expression)
[Carlyle, Kitchen et al. (2018) Journal of Proteome Research]10 -Lectures.GersteinLab.orgEMpire (Expectation Maximisation Propagation of Isoform abundance from RNA Expression)
11 -Lectures.GersteinLab.orgCumulative FractionCumulative Fraction[Carlyle, Kitchen et al. (2018) Journal of Proteome Research]Larger principal isoform dominance Less ambiguity in major isoform identification
[Carlyle, Kitchen et al. (2018) Journal of Proteome Research]12 -Lectures.GersteinLab.orgBiologically informative priors improveisoform level interpretation of MS/MS peptides,by increasing dominance of principal isoform
uORFs (Current result) Affect translation &relationship betweenprotein & RNA Feature integration to findsmall subset of uORFsthat most alter translation Future Direction:Protein v RNA using matchedsamples in the Brainspandataset single-cell data13 - Past Context:to work in the Center Quantifying the moderatestatistical correlationbetween protein & RNA PARE server EMpire (Current result) Leveraging the correlation tobetter assign peptides toisoforms EM algorithm better assignsdominant isoforms, withgreater interpretabilityLectures.GersteinLab.orgOutline: Comparing Protein & RNA Abundance
Upstream open reading frames (uORFs)may shift the expected balance between mRNA & proteinIn Battle et al. 2014 data uORF gain & loss assoc.protein level change.[McGillivray et al., NAR (‘18)]14 -uORF regulation can be affected by mutationLectures.GersteinLab.org[Zhang et al., Trends in Biochemical Sciences (‘19)]
[McGillivray et al., NAR (‘18)]15 -From a “Universe” of 1.3 Mpot. uORFsRibosome profiling experiments have low overlap inidentified uORFs.This suggests high false-negative rate, and morefunctional uORFs than currently known.Lectures.GersteinLab.orgThe population of functionaluORFs may be significant
TissueDist.ConservationInt.ATGStart[McGillivray et al., NAR (‘18)]Lectures.GersteinLab.orgAll near-cognate start codons predicted.Cross-validation on independent ribosome profiling datasets andvalidation using in vivo protein levels and ribosome occupancy inhumans (Battle et al. 2014).Expr.Level16 -Prediction & validation offunctional uORFs using 89 features
A comprehensive catalog of functional uORFs1.3M[McGillivray et al., NAR (‘18)]likely to affect translationCalibration on gold standards, suggestsgetting 70% of known17 -Predicted functional uORFs may beintersected with disease associatedvariants.180K: Large predicted positive setLectures.GersteinLab.orgUniverse ofuORFsscored via Simple Bayes algo.
uORFs (Current result) Affect translation &relationship betweenprotein & RNA Feature integration to findsmall subset of uORFsthat most alter translation Future Direction:Protein v RNA using matchedsamples in the Brainspandataset single-cell data18 - Past Context:to work in the Center Quantifying the moderatestatistical correlationbetween protein & RNA PARE server EMpire (Current result) Leveraging the correlation tobetter assign peptides toisoforms EM algorithm better assignsdominant isoforms, withgreater interpretabilityLectures.GersteinLab.orgOutline: Comparing Protein & RNA Abundance
19 -Lectures.GersteinLab.orgLeveraging New Datasets
20 -Lectures.GersteinLab.orgSchematic workflow
21 -Lectures.GersteinLab.org
22 -Sousa et al., Science 2017, 358, Pgs. 1027–1032.Lectures.GersteinLab.orgMicroRNA intervention
uORFs (Current result) Affect translation &relationship betweenprotein & RNA Feature integration to findsmall subset of uORFsthat most alter translation Future Direction:Protein v RNA using matchedsamples in the Brainspandataset single-cell data23 - Past Context:to work in the Center Quantifying the moderatestatistical correlationbetween protein & RNA PARE server EMpire (Current result) Leveraging the correlation tobetter assign peptides toisoforms EM algorithm better assignsdominant isoforms, withgreater interpretabilityLectures.GersteinLab.orgOutline: Comparing Protein & RNA Abundance
genecensus.org/expression/translatomeD Greenbaum, R Jansen, M Gersteingithub.com/rkitchen/EMpireB Carlyle, R Kitchen, J Zhang, R Wilson, T Lam, J Rozowsky,github.gersteinlab.org/uORFsP McGillivray, R Ault, M Pawashe, R Kitchen,S Balasubramanian, M GersteinBrainspan dataP Emani, T Galeev, N Sestan, A NairnLectures.GersteinLab.orgK Williams, N Sestan, M Gerstein, A Nairn24 -Acknowledgments!Proteomics.gersteinlab.org (PARE)E Yu, A Burba, M Gerstein
Outline: Comparing Protein & RNA Abundance PastContext: to work in the Center Quantifying the moderate statistical correlation between protein & RNA PARE server EMpire(Current result) Leveraging the correlation to better assign peptides to isoforms EM algorithm better
