A AA to NAA AA to EA EA to NAA B. AA to NAA ILL1 ILL2 Supplemental Figure 1. SLE patients display LFC ILL1 ILL2 ILL1 ILL2 ILL1 ILL2 ancestral expression differences. SMAD1 RPS26 LIX1L

HEBP2 0.5 (A) Limma differential expression (DE) (false discovery PSMD5-AS1 LOC105376114

ACKR1 GSTM4 FBL rate (FDR) < .05) between SLE patients of self described

CRYBB2P1 SIRPB1 TMEM106B ancestry was carried out on two cohorts from dataset PSMD5-AS1 CYP4F3 TUBB2A 3719 5198 757 PLA2G4C GSE88884: ILL1 (n = 101 AA, 540 EA, 157 NAA) and IGHV1-69-2 GTF3A

LOC90784 MIR3978 GIMAP1 ILL2 (115 AA, 577 EA, 75 NAA). The top 150 by CARS2 PADI4 RPL4P4 lowest FDR are shown for each comparison. (B-D) Venn UHRF1BP1 UHRF1BP1 RSL1D1

SULF2 diagrams depict the number of transcripts in common MFSD14B PIK3IP1

RNF19A LOC389641 TUBB2B C. AA to EA and unique between ILL1 and ILL2 for each ancestral CD93 SYNE3 EEF1A1 comparison. Both increased and decreased transcripts CIDEB MIR646HG LDHB ILL1 ILL2 RPS26P10 LOC100133331 EEF2 were included. As a control, each ancestral background

RBP7 HTATSF1P2 RPL24 was randomized into two separate groups five separate LMTK2 LOC100130353 RPL6 times and no differentially expressed genes within the ASNSD1 GLMP SNX15

DDX11L2 ROPN1L ITGA6 same ancestral background were found, even at a less

CTDSP2 LOC388572 EIF3F 1770 1875 1122 stringent FDR < 0.2. (E) Limma DE analysis of SLE ST6GALNAC2 LILRA1 RPL13 dataset GSE45291 for. 73 AA and 71 EA SLE patients UCP2 OGFRL1 EIF3L

BIN1 PPFIA1 SMIM24 with the same range of SLEDAI scores (2 – 11), similar LIMD1 DDX11L10 SRP72 0 mean SLEDAI (AA 3.78 +/- 2.46; EA 3.53 +/- 2.08) and SULF2 CXCL8 RPL18 mode of SLEDAI and DE analysis of 25 AA and 75 EA DDX11L11 UBE2Q2P2 SFMBT1

ANK3 SIGLEC12 PDXDC1 D. EA to NAA SLE patients with inactive disease (SLEDAI zero) were LINC01347 KCNE3 APEX1 ILL1 ILL2 carried out. Venn diagram depicts the number of RNVU1-19 SLC25A19 RPS3 transcripts in common and unique between the AA to EA MFSD14C LOC105369213 MRPS17

DSC2 DDX11L11 RPS13 analysis in patients with SLEDAI 2 – 11 compared to CYP4F11 SNHG17 TTC3 patients with SLEDAI 0. The complete list of DE genes SIGLEC5 LRRC37A3 PRKCSH

MIR646HG are in Supplemental Table 4. LOC100130520 EEF1D 5757 CYP4F24P NAA30 GLG1 2124 3367

PLAUR MXRA7 RPLP1P4

LOC101929819 PPT1 ASAP1-IT1

PLCG2 DHRS4-AS1 NOA1

TMC6 LOC105375566 NOD2

HSPA7 TREML4 RGPD3 TMEM91 FCAR ZC4H2 E. AA to EA GSE45291 MBIP NCF1 EIF3EP1

SNX15 TSPAN16 ZBTB6 SLEDAI SLEDAI SMIM24 IGF2R GIMAP7 2 - 11 Zero PLA2G4C LAT2 FTO -0.5 CD180 CDC16 GPR183

ANPEP OSBPL1A EIF3M

TP53I11 TUBB6 PPIL4

PCNX4 PGD ZFP36L2

ORAI2 PGDP1 EIF3FP3 427

BMX TUBB4B VSIG1 1384 298 Increased Compared to Control AA AA • NAA • EA • LDG * NAA .0625 * EA IFN A Granulocytes * B .0033 Myeloid Secrete <.0001 .0394 Mono Surface 1.0 <.0001 .0017 Monocytes Waddell.TNF.UP IL1_cytokines 0.5 Inflammasome Anti_inflammation 0.0 Tregulatory * B cells Cell Cycle -0.5 IFN Immunoglobulins Plasma Cell -1.0

Unfolded_Protein GSVA Enrichment Score SNOR Up-Low -1.5 0 20 40 60 80 100 Percent of SLE Patients with > 1SD Above Healthy Control + (30)- (18)

Decreased Compared to Control Negative+ dsDNA (21)+ dsDNA Dendritic SM SSA SSB (6) RNP RNP RNP-dsDNA+ (46) pDC Tactivated Tanergic/activated Supplemental Figure 2. Ancestral differences detected between females are similar in TCells males. (A). Percentage of male patients within each ancestry (AA n = 14, NAA n = 17, EA n = TCRA_group 93) with > (top) or < than (bottom) 1 SD GSVA enrichment scores for each cell type and TCRB_group process module. Fisher’s exact p values < .05 are indicated by different color *: pink * TCRAJ_group between AA and NAA, Dark blue * between AA and EA/NAA, red between NAA and AA/EA. TCRD_group (B) Comparison of male SLE patient IFN signature GSVA enrichment scores between five TCRG_group autoantibody groups (13 AA,16 NAA, 92 EA). n values are given in parentheses. Tukey’s Tcytotoxic multiple comparisons test was used to compare the group GSVA scores and significant p CD8T_NKT values (p < .05) are shown on the graph. The RNP+dsDNA+, RNP+dsDNA- and RNP-dsDNA NK + groups could also be positive for any one of the three other autoantibodies measured, SM, MHC II SSA or SSB. The black dotted lines represent the mean plus or minus 1 SD of the healthy Erythrocytes controls for GSVA scores. Platelets * SNOR Down-Low

0 20 40 60 80 100 Percent of Patients with < 1 SD Below Healthy Control Corticosteroids on LDGs Corticosteroids on Monocytes Corticosteroids on Anti-Inflammation 1.5 1.0 1.0 Supplemental Figure 3. Changes in GSVA scores associated with SoC drugs in dataset GSE45291. 1.0 0.5 0.5 GSVA enrichment scores for AA and EA SLE patients with SLEDAI zero or SLEDAI 2 – 11 were compared using 0.5 Tukey’s multiple comparison’s test. p values less than .05 0.0 0.0

0.0 are shown. Drug information was provided Michelle Petri. GSVA Enrichment Score GSVA Enrichment Score GSVA Enrichment Score Number of patients for each comparison are in -0.5 -0.5 -0.5 parentheses. CS (corticosteroids)

-1.0 -1.0 -1.0 AA zero CS (7) AA zero CS (7) EA zero CS (14) EA zero CS (14) AA zero Pred (7) EA 2 - 11 CS (26) EA 2 - 11 CS (26) AA 2 - 11 CS (33) AA 2 - 11 CS (33) EA zero Pred (14) EA 2 - 11 Pred (26) AA 2 - 11 Pred (33) AA zero No CS (18) AA zero No CS (18) EA Zero No CS (61) EA Zero No CS (61) AA 2 - 11 no CS (40) AA 2 - 11 no CS (40) EA 2 - 11 No CS (45) EA 2 - 11 No CS (45) AA zero No Pred (18) EA Zero No Pred (61) AA 2 - 11 no pred (40) EA 2 - 11 No Pred (45)

.03 .05 .03 /04

AZA on NK cells AZA on B cells 1.0 1.0

0.5 0.5

0.0

0.0

-0.5 GSVA Enrichment Score GSVA Enrichment Score

-0.5 -1.0

-1.5 -1.0 EA Zero No (70) EA Zero No (70) AA Zero No (23) EA 2 -11 No (65) AA Zero No (23) EA 2 -11 No (65) EA Zero AZA (5) EA Zero AZA (5) AA Zero AZA (2) AA Zero AZA (2) AA 2 - 11 No (59) AA 2 - 11 No (59) EA 2 - 11 AZA (6) EA 2 - 11 AZA (6) AA 2 - 11 AZA (15) AA 2 - 11 AZA (15) .03 <.0001 <.0001 .0004

MTX MMF on Plasma Cells MTX MMF on Ig 1.0 1.0

0.5 0.5

0.0 0.0 GSVA Enrichment Score GSVA Enrichment Score

-0.5 -0.5

-1.0 -1.0 EA Zero No (70) AA Zero No (23) EA 2 -11 No (65) EA Zero AZA (5) EA Zero No (70) AA Zero AZA (2) AA Zero No (23) EA 2 -11 No (65) EA Zero AZA (5) AA Zero AZA (2) AA 2 - 11 No (59) EA 2 - 11 AZA (6) AA 2 - 11 No (59) EA 2 - 11 AZA (6) AA 2 - 11 AZA (15) AA 2 - 11 AZA (15) EA Females Age 25 - 49 DE to Male Supplemental Figure 4. Male and female SLE patients of the same ancestry with Increased in Male similar therapy profiles have few differences in . Female SLE ILL2 patients were divided into two age groups, 25 – 49 years and > 50 years, to determine if ILL1 estrogen effects on the immune response resulted in more DE transcripts compared to 360 F 182 F males. EA females age 25 – 49 or age > 50 years from the ILL1 and ILL2 datasets were 42 M 8 51 M 146 (5 X,Y) 14 used to make SoC drug matched groups for comparison to ILL1 and ILL2 EA males (6X,Y) (1X) (Supplemental Table 15). The number of DE transcripts total and the number of transcripts on the X or Y are shown and the differentially expressed gene symbols and adjusted p values are listed in Supplemental Table 16. Increased in Female

27 80 (22 X) 25 (2X) (6X)

EA Females > Age 50 DE to Male Increased in Male ILL1 ILL2 61 F 9 81 F 13 42 M 5 (8 X,Y) 51 M (2 X,Y)

Increased in Female

17 1 (15 X) 69 (1X) (10X) Supplemental Figure 5. Changes in GSVA Plasma Cells IGS Plasma Cells in AutoAb+ IGS in Low C AutoAb+ scores associated with SoC drugs, A B C MTX/MMF- D No MTX/MMF 1.0 1.0 1.0 1.0 autoantibodies and complement (A – H) Patients of all ancestries were used for these analyses. Tukey’s multiple comparisons test was used to compare mean GSVA scores. p 0.5 0.5 0.5 0.5 values for comparisons are shown below each graph. NS means not significantly different. n values in each category are in parentheses. 0.0 0.0 0.0 0.0 Dots represent single patient scores and the error bars are mean and standard deviation. The black dotted lines represent the mean plus or minus 1 SD of the healthy controls. (A,B). GSVA Enrichment Score -0.5 -0.5 -0.5 -0.5 Female patients in GSE88884 with five autoantibody (RNP, dsDNA, SM, SSA, SSB) measurements. Plasma Cell (A) and IGS (B) -1.0 -1.0 -1.0 -1.0 GSVA scores for patients with 0/5 No Ab AutoAb+ AutoAb+ No Ab AutoAb+ AutoAb+ No Low C None CS Low C (273) (995) MTX MMF (273) (995) MTX MMF (480) (64) (415) autoantibodies (n = 273) were compared to (267) (267) (516) patients with at least 1/5 autoantibodies (n = NS < .0001 < .0001 < .0001 NS < .0001 922) and patients with at least 1/5 <.0001 <.0001 autoantibodies with concomitant MTX or MMF (n = 263). (C) The 922 patients positive for at LDG in Low C AutoAb+ NK cell least 1/5 autoantibodies were divided by the No MTX/MMF IGS in Vasculitis Plasma cells in vasculitis E F No MTX/MMF G 1.0 H 1.0 1.0 1.0 presence of low C (C3 ≤.8 g per L and/or C4 ≤ 0.1 g per L) and compared using the Welch’s t- test. (D,E) The 487 SLE patients positive for at

0.5 0.5 0.5 least one autoantibody and low C were divided 0.5 by whether they were taking corticosteroids (CS) and IGS (D) and LDG (E) signatures were determined and compared using the Welch’s t- 0.0 0.0 0.0 0.0 test. (F) NK cell GSVA scores in patients with and without 1/5 autoantibodies and taking AZA. (G,H) All patients with vasculitis were divided

GSVA Enrichment Score -0.5 by patients positive or negative for 1/5 -0.5 -0.5 -0.5 autoantibodies and IFN and plasma cell GSVA enrichment scores were compared.

-1.0 -1.0 -1.0 -1.0 No None CS + - - No AutoAb AutoAb AutoAb AutoAb (64) (415) AutoAb AutoAb+ AutoAb AutoAb AutoAb (46) Low C AutoAb (46) Low C (748) AZA (194) AZA (10) (51) (10) (51) < .0001 (248) (28) .015 .0036 .0124 .0526 < .0001 < .0001 < .0001 <.0001 .0001 .0008 SLEDAI vs AA SLEDAI vs EA Cell Cycle r2 = .03 SLEDAI vs NAA IFN r2 = .03 2 2 IFN r = .05 Ig Cells r = .03 Cell cycle r2 = .03 2 2 Plasma Cell r = .03 Plasma Cells r = .03 IFN r2 = .03 2 Ig Chains r = .02 LDG r2 = .03 LDG r2 = .03 0.5 Cell Cycle r2 = .05 0.5 0.5 Mye Secrete r2 = .03 Plasma cells r2 = .03 Ig Chains r2 = .03 TNF r2 = .03 TNF r2 = .03

0.0 0.0 0.0 10 20 30 10 20 30 40 50 10 20 30 SLEDAI SLEDAI SLEDAI GSVA Enrichment Score GSVA Enrichment Score GSVA Enrichment Score -0.5 -0.5 -0.5

dsDNA titers vs GSVA AA IFN r2 = .03 dsDNA titers vs GSVA EA IFN r2 = .1 dsDNA titers vs GSVA NAA LDG r2 = .06 Cell cycle r2 = .07 Cell cycle r2 = .12 TNF r2 = .03 1.0 1.0 1.0 LDG r2 = .03 LDG r2 = .05 Cell cycle r2 = .12 Ig Chains r2 = .03 Ig Chains r2 = .08 PC r2 = .06 PC r2 = .04 PC r2 = .09 IFN r2 = .11 0.5 0.5 Mye Secrete r2 = .06 0.5 Ig Chains r2 = .05 TNF r2 = .07 Mye Secrete r2 = .05

0.0 0.0 0.0 100 200 300 400 100 200 300 400 100 200 300 400

-0.5 -0.5 -0.5 GSVA Enrichment Score GSVA Enrichment Score GSVA Enrichment Score dsDNA_IU

-1.0 dsDNA_IU -1.0 -1.0 dsDNA_IU

Mye Secrete r2 = .04 C3 vs GSVA AA C3 vs GSVA EA TNF r2 = .04 C3 vs GSVA NAA 2 PC r2 = .08 TNF r = .05 Mye Secrete r2 = .05 Mye Secrete r2 = .05 1.0 TNF r2 = .04 1.0 1.0 Cell cycle r2 = .16 2 UPR r2 = .06 Cell cycle r = .12 Ig Chains r2 = .11 2 Treg r2 = .07 Ig Chains r = .11 PC r2 = .12 2 Ig Chains r2 = .08 PC r = .11 UPR r2 = .06 2 0.5 LDG r2 = .03 0.5 0.5 UPR r = .03 Treg r2 = .07 2 IFN r2 = .07 Treg r = .05 IFN r2 = .13 2 Cell cycle r2 = .11 IFN r = .15

0.0 0.0 0.0

0.5 1.0 1.5 2.0 2.5 0.5 1.0 1.5 2.0 2.5 0.5 1.0 1.5 2.0

-0.5 -0.5 -0.5 GSVA Enrichment Score GSVA Enrichment Score GSVA Enrichment Score

-1.0 -1.0 -1.0 C3_GperL C3_GperL C3_GperL

Supplemental Figure 6. Linear regression analysis between SLEDAI values, dsDNA titers (IU), and C3 gram/Liter (GperL) values and 34 cell and process module GSVA enrichment scores for female AA SLE patients (n = 208), EA SLE patients (n = 1100) and NAA SLE patients (n = 227). Only r2 values for GSVA enrichment categories with p < .05 are shown. A C

B Supplemental Figure 7. A machine learning approach predicted AA from a mixture of AA, EA and NAA SLE Model AUC Accuracy Sensivity Specificity Kappa patients and demonstrated the perturbed B cell axis in Type AA SLE. (A) SLE patients were classified as African American (AA) using logistic regression, generalized linear Logisc 0.93 0.92 0.84 0.93 0.70 models (GLM), and support vector machine (SVM) Regression classifiers. ROC curve for logistic regression and the two different machine learning models in GSE88884 (ILL1 and GLM 0.97 0.95 0.78 0.98 0.80 ILL2 combined). (B) Table showing the metrics of three machine learning models. (C) Top 25 predictors SVM 0.90 0.96 0.82 0.98 0.83 determined by SVM model. GSE88884 Supplemental Figure 8. Sex module scores for SLE XIST + TSIX - (UTY + RPS4Y1 + USP9Y) patients listed as males or females and unknown controls in microarray dataset GSE88884. Log2 10 expression values were used to compute sex module scores using the formula Sex = XIST + TSIX – (UTY + 5 RPS4Y1 + USP9Y). 5 SLE patients (3 male and 2 female) with reported sex in GSE88884 ILL1 dataset were found to have expression of genes consistent with 0 the opposite sex.

-5

-10 GSVA Enrichment Scores

-15

-20

SLE Males (121) SLE Females (1566)

Unknown Sex GSE88884 CTLs