Presentation is loading. Please wait.

Presentation is loading. Please wait.

SJIA flare signature analysis 2 Discovery set (ND.SAF vs. AF.QOM.RD.KD.FI) 1 LCMS raw spectra Peak finding peak alignment feature extraction Urine peptide.

Similar presentations


Presentation on theme: "SJIA flare signature analysis 2 Discovery set (ND.SAF vs. AF.QOM.RD.KD.FI) 1 LCMS raw spectra Peak finding peak alignment feature extraction Urine peptide."— Presentation transcript:

1 SJIA flare signature analysis 2 Discovery set (ND.SAF vs. AF.QOM.RD.KD.FI) 1 LCMS raw spectra Peak finding peak alignment feature extraction Urine peptide index NSC feature selection Ten-fold Cross-validation Urine profilingUrine biomarker analysis (NSC, LDA, ROC) Feature selection Identification of FGA peptides LDA analysis 6 peptide biomarker panel ROC analysis 500 bootstrap samples 3 Classification ND.SAF vs. KD.FI ND.SAF vs. QOM.RD SJIA flare signature analysis 4 Discovery set (SAF, QOM) NSC feature selection Ten-fold Cross-validation Feature selection LDA analysis 6 protein biomarker panel ROC analysis 5 Classification Training Samples SAF vs. QOM “bootstrap” samples SAF vs. QOM Literature review Antibody array construction Hypothesis generation Cherry pick 43 protein antibodies Antibody array assay Antibody array profiling Plasma biomarker analysis (NSC, LDA, ROC) 6 FIGURE 1

2 TABLE 1

3 TABLE 2

4 FGA(20-35)1536.61ADSGEGDFLAEGGGVR FGA(607-622)1639.77 AGSEADHEGTHSTKRG FGA(605-621)1826.80 DEAGSEADHEGTHSTKR FGA(605-622)1883.80DEAGSEADHEGTHSTKRG FGA(605-628)2560.2DEAGSEADHEGTHSTKRGHAKSRP FGA(605-629)*2659.24DEAGSEADHEGTHSTKRGHAKSRPV MH+ProteinSequence Relative abundance ND.SAF AF.QOM.RD.KD.FI 0.0015 0.0986 0.3225 0.6016 0.0037 0.1443 -0.0004 -0.024 -0.0787 -0.1467 -0.0009 -0.0352 TABLE 3

5 FGA(605-621) 1826.80 FGA(605-622) 1883.80 FGA(605-629)* 2659.24 SAF AF QOM RDKDFIHCND FGA(607-622) 1639.77 FGA(605-628) 2560.2 FGA(20-38) 1536.69 SAF AF QOM RDKDFIHCND FGA peptide marker distribution Relative abundance FIGURE 2

6 Classification 2+1823+23 Clinical diagnosis ND.SAFKD.FI n = LDA 141 645 Predicted as SJIA F Predicted as non SJIA F Percent Agreement with clinical diagnosis 70%97.8% +- 89.4% Overall P = 6.7X10 -9 AB ND SAFKD FI Predicted probabilities Patient samples Sensitivity 1- Specificity Mean(AUC): 95.6% C FIGURE 3

7 Classification 2+1818+9 Clinical diagnosis ND.SAFQOM.RD n = LDA 132 725 Predicted as SJIA F Predicted As SJIA Q Percent Agreement with clinical diagnosis 65%92.6% +- 80.9% Overall P = 6.1X10 -5 AB ND SAFQOM RD Predicted probabilities Patient samples Sensitivity 1- Specificity Mean(AUC): 91.0% C FIGURE 4

8 MH+SequenceFGA location 1465.58 DSGEGDFLAEGGGVR21-35 1536.61 ADSGEGDFLAEGGGVR20-35 1775.74 DSGEGDFLAEGGGVRGPR21-38 1846.77 ADSGEGDFLAEGGGVRGPR20-38 1339.68SQLQKVPPEWK239-249 1252.65 QLQKVPPEWK240-249 1686.71GGSTSYGTGSETESPRN272-288 2473.02GGSTSYGTGSETESPRNPSSAGSWN272-296 2375.07GSTGNRNPGSSGTGGTATWKPGSSGP303-328 1360.61 PGSSGTGGTATWKPG310-324 1554.66DGFRHRHPDEAAF507-519 2553.01SSSYSKQFTSSTSYNRGDSTFES576-598 2768.13SSSYSKQFTSSTSYNRGDSTFESKS576-600 2931.15SSSYSKQFTSSTSYNRGDSTFESKSY576-601 1913.75 QFTSSTSYNRGDSTFES582-598 1354.52 DEAGSEADHEGTH605-617 1542.60 DEAGSEADHEGTHST605-619 2020.90 DEAGSEADHEGTHSTKRGH605-623 2091.92 DEAGSEADHEGTHSTKRGHA605-624 2560.20 DEAGSEADHEGTHSTKRGHAKSRP605-628 2659.24 DEAGSEADHEGTHSTKRGHAKSRPV605-629 2344.12 GSEADHEGTHSTKRGHAKSRPV608-629 2730.26 ADEAGSEADHEGTHSTKRGHAKSRPV604-629 2293.96 MADEAGSEADHEGTHSTKRGHA603-624 2762.25 MADEAGSEADHEGTHSTKRGHAKSRP603-628 2861.28 MADEAGSEADHEGTHSTKRGHAKSRPV603-629 2877.31 MADEAGSEADHEGTHSTKRGHAKSRPV*603-629 2122.81 SYKMADEAGSEADHEGTHST600-619 2672.11 SYKMADEAGSEADHEGTHSTKRGHA600-624 3239.46 SYKMADEAGSEADHEGTHSTKRGHAKSRPV600-629 3255.46 SYKMADEAGSEADHEGTHSTKRGHAKSRPV*600-629 851.49 GHAKSRPV622-629 I II III IV V VI VII TABLE 4

9 B D TrainingBootstrapping testing A FIGURE 5 SAFQOM Predicted probabilities Patient samples SAFQOM Training samples n = 39 2514 Clinical diagnosis FQ n = LDA 234 210 Classified as F Classified as Q Percent Agreement with clinical diagnosis 92%71.4% +- 84.6% Overall P = 7.9X 10 -5 Bootstrapping samples n = 52 4111 Clinical diagnosis FQ n = Testing 362 59 Classified as F Classified as Q Percent Agreement with clinical diagnosis 87.8%81.8% +- 86.5% Overall P = 2.4 X 10 -5 C SJIA Relative abundance biomarkersSAFQOM TIMP-10.2782-0.4967 IL-180.1735-0.3099 RANTES0.1681-0.3002 P-Selectin0.1616-0.2885 MMP-90.1308-0.2335 L-Selectin0.0121-0.0216

10 Sensitivity 1- Specificity Mean(AUC): 92.2% Sensitivity 1- Specificity Mean(AUC): 90.7% TrainingBootstrapping testing FIGURE 6

11 A C TrainingBootstrapping testing FIGURE 5 SAFQOM Predicted probabilities Patient samples SAFQOM Training samples n = 47 LDA 132 725 Classified as F Classified as Q Percent Agreement with clinical diagnosis 65%92.6% +- 80.9% Overall P = 6.1X 10 -5 Bootstrapping samples n = 24 159 Clinical diagnosis FQ n = Testing 71 88 Classified as F Classified as Q Percent Agreement with clinical diagnosis 87.8%81.8% +- 86.5% Overall P = 2.4 X 10 -5 B SJIA Clinical diagnosis ND.SAFQOM.RD n = SJIA 2+1818+9

12 Sensitivity 1- Specificity Mean(AUC): 90.8% Sensitivity 1- Specificity Mean(AUC): 80.9% TrainingBootstrapping testing FIGURE 6


Download ppt "SJIA flare signature analysis 2 Discovery set (ND.SAF vs. AF.QOM.RD.KD.FI) 1 LCMS raw spectra Peak finding peak alignment feature extraction Urine peptide."

Similar presentations


Ads by Google