Methods and systems for determining proportions of distinct cell subsets

1 . A method for treating a subject for cancer, comprising:

(a) assaying a biological sample comprising peripheral blood mononuclear cells from a subject having cancer, thereby generating a feature profile m, wherein the biological sample comprises a first plurality of distinct leukocyte cell subsets, wherein the cancer comprises brain cancer, blood cancer, or non-brain solid tumor cancer,

wherein the feature profile m comprises combinations of features associated with the first plurality of distinct leukocyte cell subsets,

wherein the feature profile m comprises a gene expression profile of leukocyte cells in the biological sample, wherein the gene expression profile represents a ribonucleic acid (RNA) transcriptome of the leukocyte cells in the biological sample:

(b) optimizing, by a computer processor, a regression between the feature profile m and a reference matrix B of feature signatures for a second plurality of distinct leukocyte cell subsets in the biological sample, wherein the feature profile m is modeled as a linear combination of the reference matrix B,

wherein the reference matrix B comprises an LM22 signature matrix of relative gene expression values across 22 leukocyte subsets, wherein the LM22 signature matrix comprises ABCB4, ABCB9, ACAP1, ACHE, ACP5, ADAM28, ADAMDEC1, ADAMTS3, ADRB2, AIF1, AIM2, ALOX15, ALOX5, AMPD 1 , ANGPT4, ANKRD55, APOBEC3A, APOBEC3G, APOL3, APOL6, AQP 9 , ARHGAP22, ARRB1, ASGR1, ASGR2, ATHL1, ATP8B4, ATXN80S, AZU1, BACH2, BANK1, BARX2, BCL11B, BCL2A1, BCL7A, BEND5, BFSP1, BHLHE41, BIRC3, BLK, BMP2K, BPI, BRAF, BRSK2, BST1, BTNL8, C11orf80, Clorf54, C3AR1, C5AR1, C5AR2, CA8, CAMP, CASP5, CCDC102B, CCL1, CCL13, CCL14, CCL17, CCL18, CCL19, CCL20, CCL22, CCL23, CCL4, CCL5, CCL7, CCL8, CCND2, CCR10, CCR2, CCR3, CCR5, CCR6, CCR7, CD160, CD180, CD19, CD1A, CD1B, CD1C, CD1D, CD1E, CD2, CD209, CD22, CD244, CD247, CD27, CD28, CD300A, CD33, CD37, CD38, CD3D, CD3E, CD3G, CD4, CD40, CD40LG, CD5, CD6, CD68, CD69, CD7, CD70, CD72, CD79A, CD79B, CD80, CD86, CD8A, CD8B, CD96, CDA, CDC25A, CDH12, CDHR1, CDK6, CEACAM3, CEACAM8, CEMP1, CFP, CHI3L1, CHI3L2, CHST15, CHST7, CLC, CLCA3P, CLEC10A, CLEC2D, CLEC4A, CLEC7A, CLIC2, CMA1, COL8A2, COLO, CPA3, CR2, CREB5, CRISP3, CRTAM, CRYBB1, CSF1, CSF2, CSF3R, CST7, CTLA4, CTSG, CTSW, CXCL10, CXCL11, CXCL13, CXCL3, CXCL5, CXCL9, CXCR1, CXCR2, CXCR5, CXCR6, CXorf57, CYP27A1, CYP27B1, DACH1, DAPK2, DCSTAMP, DEFA4, DENNDSB, DEPDC5, DGKA, DHRS11, DHX58, DPEP2, DPP4, DSC1, DUSP2, EAF2, EBI3, EFNA5, EGR2, ELANE, EMR1, EMR2, EMR3, EPB41, EPHA1, EPN2, ETS1, ETV3, FAIM3, FAM124B, FAM174B, FAM198B, FAM212B, FAM65B, FASLG, FBXL8, FCERIA, FCER2, FCGR2B, FCGR3B, FCN1, FCRL2, FES, FFAR2, FLJ13197, FLT3LG, FLVCR2, FOSB, FOXP3, FPR1, FPR2, FPR3, FRK, FRMD4A, FRMD8, FZD2, FZD3, GAL3ST4, GALR1, GFI1, GGT5, GIPR, GNG7, GNLY, GPC4, GPR1, GPR171, GPR18, GPR183, GPR19, GPR25, GPR65, GPR97, GRAP2, GSTT1, GUSBP11, GYPE, GZMA, GZMB, GZMH, GZMK, GZMM, HAL, HCK, HDC, HESX1, HHEX, HIC1, HISTIH2AE, HIST1H2BG, HK3, HLA-DOB, HLA-DOA1, HMGB3P30, HNMT, HOXA1, HPGDS, HPSE, HRH1, HSPA6, HTR2B, ICA1, ICOS, IDO1, IFI44L, IFNA10, IFNG, IGHD, IGHE, IGHM, IGKC, IGLL3P, IGSF6, IL12B, IL12RB2, IL17A, IL18R 1 , IL18RAP, IL1A, IL1B, IL1RL1, IL21, IL26, IL2RA, IL2RB, IL3, IL4, IL4R, IL5, ILSRA, IL7, IL7R, IL9, IRF8, ITK, KCNA3, KCNG2, KIAA0226L, KIAA0754, KIR2DL1, KIR2DL4, KIR2DS4, KIR3DL2, KIRREL, KLRB1, KLRC3, KLRC4, KLRD1, KLRF1, KLRG1, KLRK1, KRT18P50, KYNU, LAG3, LAIR2, LAMP3, LAT, LCK, LEF1, LHCGR, LILRA2, LILRA3, LILRA4, LILRB2, LIME1, LINC00597, LINC00921, LOC100130100, LOC126987, LRMP, LST1, LTA, LTB, LTC4S, LY86, LY9, MAGEA11, MAK, MANlA1, MANEA, MAP3K13, MAP4K1, MAP4K2, MAP9, MARCH3, MARCO, MAST1, MBL2, MEFV, MEP1A, MGAM, MICAL3, MMP12, MMP25, MMP9, MNDA, MROH7, MS4A1, MS4A2, MS4A3, MS4A6A, MSC, MXD1, MYB, MZB1, NAALADL1, NCF2, NCR3, NFE2, NIPSNAP3B, NKG7, NLRP3, NMBR, NME8, NOD2, NOX3, NPAS1, NPIPB15, NPL, NR4A3, NTN3, NTRK1, ORC1, OSM, P2RX1, P2RX5, P2RY10, P2RY13, P2RY14, P2RY2, PADI4, PAQR5, PASK, PAX7, PBXIP1, PCDHA5, PDCD1, PDCD 1 LG2, PDE6C, PDK1, PGLYRP1, PIK3IP1, PKD2L2, PLAIA, PLA2G 7 , PLCH2, PLEKHF1, PLEKHG3, PMCH, PNOC, PPBP, PPFIBP1, PRF1, PRG2, PRR5L, PSG2, PTGDR, PTGER2, PTG1R, PTPRCAP, PTPRG, PVRIG, QPCT, RAB27B, RALGPS2, RASA3, RASGRP2, RASGRP3, RASSF4, RCAN3, REN, RENBP, REPS2, RGS1, RGS13, RNASE2, RNASE6, RPL10L, RPL3P7, RRP12, RRP9, RSAD2, RYR1, S100A12, S1PR5, SAMSN1, SCN9A, SEC31B, SELL, SEPT5, SEPT8, SERGEF, SH2D1A, SIGLEC1, SIK1, SIRPG, SIT1, SKA1, SKAP1, SLAMF1, SLAMF8, SLC12A1, SLC12A8, SLC15A3, SLC2A6, SLC7A10, SLCO5A1, SMPD3, SMPDL3B, SOCS1, SP140, SPAG4, SPIB, SPOCK2, SSX1, ST3GAL6, ST6GALNAC4, ST8SIA1, STAP1, STEAP4, STXBP6, TARDBPP1, TBX21, TCF7, TCL1A, TEC, TEP1, TGM5, TLR2, TLR7, TLR8, TMEM156, TMEM255A, TNFAIP6, TNFRSF10C, TNFRSF11A, TNFRSF13B, TNFRSF17, TNFRSF4, TNFSF14, TNIP3, TPSAB1, TRAC, TRAF4, TRAT1, TRAV12-2, TRAV13-1, TRAV13-2, TRAV21, TRA V8-6, TRA V9-2, TRBC1, TRDC, TREM1, TREM2, TREML2, TRIB2, TRPM4, TRPM6, TSHR, TTC38, TXK, TYR, UBASH3A, UGT1A8, UGT2B17, UPK3A, VILL, VNN1, VNN2, VNN3, VPREB3, WNT5B, WNT7A, ZAP70, ZBP1, ZBTB10, ZBTB32, ZFP36L2, ZNF135, ZNF165, ZNF204P, ZNF222, ZNF286A, ZNF324, and ZNF442, and

wherein the optimizing comprises solving for a set of regression coefficients f of the regression, wherein the solving minimizes a linear loss function and an L2-norm penalty function:

(c) estimating a relative proportion of one or more distinct leukocyte cell subsets among the second plurality of distinct leukocyte cell subsets present in the biological sample, based at least in part on the set of regression coefficients f;

(d) predicting a clinical outcome of the cancer, based at least in part on a comparison between the estimated relative proportions of the one or more distinct leukocyte cell subsets present in the biological sample and a pre-determined association of the one or more distinct leukocyte cell subsets with clinical outcomes for the cancer, wherein the clinical outcome comprises survival of the subject; and

(e) based on the predicted clinical outcome of the cancer, administering a cancer therapy to the subject thereby treating the cancer of the subject, wherein the cancer therapy comprises a member selected from the group consisting of a chemotherapy, an immunotherapy, and an immunochemotherapy.

2 . The method of claim 1 , wherein the solving for the set of regression coefficients f further comprises selecting a subset of features in the reference matrix B among a plurality of different subsets of feature signatures of the reference matrix B to minimize the linear loss function.

3 . The method of claim 1 , wherein the linear loss function is a linear ε-insensitive loss function.

4 . The method of claim 1 , wherein the optimizing further comprises using support vector regression (SVR).

5 . The method of claim 4 , wherein the support vector regression is ε-SVR.

6 . The method of claim 4 , wherein the support vector regression is v(nu)-SVR.

7 . The method of claim 1 , further comprising determining a significance value for the estimating.

8 . The method of claim 7 , wherein determining the significance value comprises:

(i) generating a random feature profile m* comprising features randomly selected from a parent feature profile, wherein the parent feature profile comprises the feature profile m, and wherein the feature profile m and the random feature profile m* have the same Euclidean norm;

(ii) optimizing a second regression between the random feature profile m* and the reference matrix B, wherein the random feature profile m* is modeled as a linear combination of the reference matrix B,

wherein the optimizing in (ii) comprises solving for a set of regression coefficients f* of the second regression, wherein the solving for the set of regression coefficients f* of the second regression minimizes a linear loss function and an L2-norm penalty function;

(iii) calculating a product of the set of regression coefficients f* and the reference matrix B to generate a reconstituted feature profile;

(iv) determining a difference measurement between the random feature profile m* and the reconstituted feature profile; and

(v) determining the significance value based at least in part on a distribution of difference measurements determined from performing a plurality of i iterations of (i) to (iv).

9 . The method of claim 1 , wherein the reference matrix B comprises at least one distinct feature that is present in the feature profile m of two or more distinct leukocyte cell subsets of the second plurality of distinct leukocyte cell subsets.

10 . The method of claim 1 , wherein the plurality of distinct leukocyte cell subsets comprises two or more distinct immune cell types.

11 . The method of claim 1 , wherein a non-negative regression coefficient of the set of regression coefficients f is indicative of a relative proportion of a corresponding distinct cell subset among the second plurality of distinct cell subsets present in the biological sample.

12 . The method of claim 11 , further comprising setting negative regression coefficients of the set of regression coefficients f to zero values, and normalizing the non-negative regression coefficients of the set of regression coefficients f, thereby generating the estimated relative proportion of the one or more distinct cell subsets.

13 . A method for treating a subject for cancer, comprising:

administering a cancer therapy to the subject thereby treating the cancer of the subject, based on a predicted clinical outcome of the cancer, wherein the cancer comprises brain cancer, blood cancer, or non-brain solid tumor cancer, and wherein the clinical outcome comprises survival of the subject therapy;

wherein the cancer therapy comprises a member selected from the group consisting of a chemotherapy, an immunotherapy, and an immunochemotherapy; and

wherein the predicted clinical outcome of the cancer is determined at least in part by:

(a) assaying a biological sample comprising peripheral blood mononuclear cells from the subject, thereby generating a feature profile m, wherein the biological sample comprises a first plurality of distinct leukocyte cell subsets,

wherein the feature profile m comprises combinations of features associated with the first plurality of distinct leukocyte cell subsets,

wherein the feature profile m comprises a gene expression profile of leukocyte cells in the biological sample, wherein the gene expression profile represents a ribonucleic acid (RNA) transcriptome of the leukocyte cells in the biological sample;

(b) optimizing, by a computer processor, a regression between the feature profile m and a reference matrix B of feature signatures for a second plurality of distinct leukocyte cell subsets in the biological sample, wherein the feature profile m is modeled as a linear combination of the reference matrix B,

wherein the reference matrix B comprises an LM22 signature matrix of relative gene expression values across 22 leukocyte subsets, wherein the LM22 signature matrix comprises ABCB4, ABCB9, ACAP1, ACHE, ACP5, ADAM28, ADAMDEC1, ADAMTS3, ADRB2, AIF1, AIM2, ALOX15, ALOX5, AMPD1, ANGPT4, ANKRD55, APOBEC3A, APOBEC3G, APOL3, APOL6, AQP 9 , ARHGAP22, ARRB1, ASGR1, ASGR2, ATHL1, ATP8B4, ATXN8OS, AZUL, BACH2, BANK1, BARX2, BCL11B, BCL2A1, BCL7A, BENDS, BFSP1, BHLHE41, BIRC3, BLK, BMP2K, BPI, BRAF, BRSK2, BST1, BTNL8, C11orf80, Clorf54, C3AR1, C5AR1, C5AR2, CA8, CAMP, CASP5, CCDCl02B, CCL1, CCL13, CCL14, CCL17, CCL18, CCL19, CCL20, CCL22, CCL23, CCL4, CCL5, CCL7, CCL8, CCND2, CCR10, CCR2, CCR3, CCR5, CCR6, CCR7, CD160, CD180, CD19, CD1A, CD1B, CD1C, CD1D, CD1E, CD2, CD209, CD22, CD244, CD247, CD27, CD28, CD300A, CD33, CD37, CD38, CD3D, CD3E, CD3G, CD4, CD40, CD40LG, CD5, CD6, CD68, CD69, CD7, CD70, CD72, CD79A, CD79B, CD80, CD86, CD8A, CD8B, CD96, CDA, CDC25A, CDH12, CDHR1, CDK6, CEACAM3, CEACAM8, CEMP1, CFP, CHI3L1, CHI3L2, CHST15, CHST7, CLC, CLCA3P, CLEC10A, CLEC2D, CLEC4A, CLEC7A, CLIC2, CMA1, COL8A2, COLQ, CPA3, CR2, CREB5, CRISP3, CRTAM, CRYBB1, CSF1, CSF2, CSF3R, CST7, CTLA4, CTSG, CTSW, CXCL10, CXCL11, CXCL13, CXCL3, CXCL5, CXCL9, CXCR1, CXCR2, CXCR5, CXCR6, CXorf57, CYP27A1, CYP27B1, DACH1, DAPK2, DCSTAMP, DEFA4, DENND5B, DEPDC5, DGKA, DHRS11, DHX58, DPEP2, DPP4, DSC1, DUSP2, EAF2, EBI3, EFNA5, EGR2, ELANE, EMR1, EMR2, EMR3, EPB41, EPHA1, EPN2,ETS1. ETV3. FAIM3. FAM124B. FAM174B. FAM198B. FAM212B, FAM65B. FASLG, FBXL8, FCERIA, FCER2. FCGR2B. FCGR3B. FCN1. FCRL2. FES. FFAR2. FLJ13197. FLT3LG.FLVCR2, FOSB. FOXP3. FPR1. FPR2. FPR3. FRK. FRMD4A. FRMD8. FZD2, FZD3, GAL3ST4. GALR1. GFI1. GGTS. GIPR. GNG7. GNLY. GPC4. GPR1. GPR171. GPR18.GPR183, GPR19. GPR25. GPR65. GPR97. GRAP2. GSTT1, GUSBP11. GYPE. GZMA. GZMB, GZMH. GZMK, GZMM. HAL. HCK, HDC, HESX1. HHEX, HIC1. HIST1H2AE. HIST1H2BG, HK3. HLA-DOB, HLA-DOA1. HMGB3P30. HNMT. HOXAl. HPGDS. HPSE. HRH1, HSPA6.HTR2B. ICA1. ICOS, IDO1. IFI44L, IFNA10. IFNG. IGHD, IGHE, IGHM. IGKC. IGLL3P.IGSF6. IL12B. IL12RB2. IL17A, IL18R1. IL18RAP. IL1A, IL1B, IL1RL1, IL21, IL26. IL2RA.IL2RB. IL3. IL4. IL4R. ILS, ILSRA. IL7, IL7R. IL9, IRF8, ITK. KCNA3, KCNG2. KIAA0226L, KIAA0754. KIR2DL1, KIR2DL4, KIR2DS4, KIR3DL2. KIRREL, KLRB1. KLRC3, KLRC4.KLRD1. KLRF1, KLRG1. KLRK1. KRT18P50. KYNU. LAG3. LAIR2. LAMP3. LAT. LCK, LEFi. LHCGR. LILRA2. LILRA3. LILRA4, LILRB2. LIMEl. LINC00597, LINC00921, LOC100130100. LoC126987. LRMP. LST1, LTA, LTB. LTC4S. LY86, LY9. MAGEAll, MAK, MANlAl. MANEA. MAP3K13, MAP4K1. MAP4K2. MAP9. MARCH3. MARCO. MAST1. MBL2. MEFV, MEP1A. MGAM. MICAL3. MMP12. MMP25. MMP9, MNDA. MROH7. MS4A1.MS4A2. MS4A3. MS4A6A. MSC, MXD1. MYB. MZB1. NAALADL1. NCF2. NCR3. NFE2.NIPSNAP3B, NKG7, NLRP3. NMBR. NME8. NOD2. NOX3. NPAS1. NPIPB15. NPL. NR4A3. NTN3, NTRK1. ORCi. OSM. P2RX1. P2RX5, P2RY10. P2RY13. P2RY14. P2RY2. PADI4. PAQRS. PASK, PAX7, PBXIP1. PCDHAS, PDCD1. PDCD1LG2. PDE6C. PDK1. PGLYRP1. PIK3IP1. PKD2L2. PLAIA, PLA2G7. PLCH2. PLEKHF1. PLEKHG3, PMCH. PNOC. PPBP, PPFIBP1. PRF1. PRG2. PRRSL. PSG2. PTGDR. PTGER2. PTG1R, PTPRCAP. PTPRG. PVRIG, QPCT, RAB27B. RALGPS2. RASA3. RASGRP2, RASGRP3. RASSF4, RCAN3. REN. RENBP.REPS2. RGS1. RGS13. RNASE2. RNASE6, RPLIOL. RPL3P7, RRP12. RRP9. RSAD2, RYRi. S100A12. SiPR5. SAMSN1. SCN9A, SEC31B. SELL. SEPTS, SEPT8. SERGEF, SH2D1A. SIGLECI. SIKi. SIRPG, SITE, SKAL. SKAPI. SLAMFI. SLAMF8. SLC12A1. SLC12A8. SLC15A3. SLC2A6. SLC7A10. SLCO5A1. SMPD3. SMPDL3B. SOCS1. SP140. SPAG4. SPIB. SPOCK2, SSX1. ST3GAL6. ST6GALNAC4. ST8SIA1. STAPl. STEAP4. STXBP6. TARDBPP1. TBX21. TCF7. TCL1A. TEC. TEPi. TGMS. TLR2. TLR7. TLR8. TMEM156. TMEM255A. TNFAIP6, TNFRSF1OC. TNFRSF11A, TNFRSF13B. TNFRSF17. TNFRSF4. TNFSF14. TNIP3. TPSAB1. TRAC, TRAF4. TRATl. TRAV12-2, TRAV13-1, TRAV13-2, TRAV21. TRAV8-6, TRAV9-2, TRBC1, TRDC, TREM1, TREM2, TREML2, TRIB2, TRPM4, TRPM6, TSHR, TTC38, TXK, TYR, UBASH3A, UGT1A8, UGT2B17, UPK3A, VILL, VNN1, VNN2, VNN3, VPREB3, WNT5B, WNT7A, ZAP70, ZBP1, ZBTB10, ZBTB32, ZFP36L2, ZNF135, ZNF165, ZNF204P, ZNF222, ZNF286A, ZNF324, and ZNF442, and

wherein the optimizing comprises solving for a set of regression coefficients f of the regression, wherein the solving minimizes a linear loss function and an L2-norm penalty function;

(c) estimating a relative proportion of one or more distinct leukocyte cell subsets among the second plurality of distinct leukocyte cell subsets present in the biological sample, based at least in part on the set of regression coefficients f; and

(d) predicting a clinical outcome of the cancer, based at least in part on a comparison between the estimated relative proportions of the one or more distinct leukocyte cell subsets present in the biological sample and a pre-determined association of the one or more distinct leukocyte cell subsets with clinical outcomes for the cancer.

14 . The method of claim 13 , wherein the solving for the set of regression coefficients f further comprises selecting a subset of features in the reference matrix B among a plurality of different subsets of feature signatures of the reference matrix B to minimize the linear loss function.

15 . The method of claim 13 , wherein the linear loss function is a linear E-insensitive loss function.

16 . The method of claim 13 , wherein the optimizing further comprises using support vector regression (SVR).

17 . The method of claim 16 , wherein the support vector regression is ε-SVR.

18 . The method of claim 16 , wherein the support vector regression is v(nu)-SVR.

19 . The method of claim 13 , further comprising determining a significance value for the estimating.

20 . The method of claim 19 , wherein determining the significance value comprises:

(i) generating a random feature profile m* comprising features randomly selected from a parent feature profile, wherein the parent feature profile comprises the feature profile m, and wherein the feature profile m and the random feature profile m* have the same Euclidean norm;

(ii) optimizing a second regression between the random feature profile m* and the reference matrix B, wherein the random feature profile m* is modeled as a linear combination of the reference matrix B,

wherein the optimizing in (ii) comprises solving for a set of regression coefficients f* of the second regression, wherein the solving for the set of regression coefficients f* of the second regression minimizes a linear loss function and an L2-norm penalty function;

(iii) calculating a product of the set of regression coefficients f* and the reference matrix B to generate a reconstituted feature profile;

(iv) determining a difference measurement between the random feature profile m* and the reconstituted feature profile; and

(v) determining the significance value based at least in part on a distribution of difference measurements determined from performing a plurality of i iterations of (i) to (iv).

21 . The method of claim 13 , wherein a non-negative regression coefficient of the set of regression coefficients f is indicative of a relative proportion of a corresponding distinct cell subset among the second plurality of distinct cell subsets present in the biological sample.