Huiyi - POSTRE Project

Author

Huiyi Hu

Published

December 12, 2023

POSTRE (Prediction Of STRuctural variant Effects)

A bioinformatics tool to predict the pathological effects of human structural variants in congenital diseases

POSTRE manuscript: https://academic.oup.com/nar/article/51/9/e54/7097664

Github: https://github.com/vicsanga/Postre

Software Introduction

Input:

  1. Type of SV (Deletions, Duplications, Inversions, Insertions, Translocations)
  2. The genomic coordinates of the SV breakpoints: Allow POSTRE to determine candidate genes (causative and non-causative genes)
  3. Type of Congenital abnormality

Calculating Pathogenicity Score (PS Score)….

  1. Consider the phenotype/s associated with the SV in order to select cell types/tissues relavent for such phenotype/s
  2. The SV breakpoints are mapped in the context of TADs identified in the selected cell type/tissue
  3. All the genes located within TADs are selected as potential candidates associated with the patient diseases (candidate genes)
  4. All candidate genes are ranked according to their PS score in each of the considered cell types/tissues

Features used to calculate the gene PS:

genePhenoScore :

  • Quantification of the relationship between the candidate genes and the SV associated phenotype

  • genePhenoscore = 1 if the candidate gene is associated with the same phenotype in human

  • genePhenoscore = 0.5 if the candidate gene is associated with the same phenotype in mice but not in human

  • genePhenoscore = 0 Otherwise

geneExpressionScore:

  • Particularly relevant for loss of function (LOF) situation, as the disease-causative genes must be expressed in the relevant cell types/tissues (for both”coding” and “long-range” cases)

  • geneExpressionScore = 1 if candidate gene with FPKM ≥ 10

  • geneExpressionScore = 0 if candidate gene with FPKM ≤ 1

  • 0-1 min-max normalization is used for candidate gene with 1 < FPKM <10

  • *FPKM = Fragments Per Kilobase of transcript per Million mapped reads

geneEnhancerScore:

  • Quantification of the changes in enhancer activity between Control and Patient alleles

  • LOF scenario: geneEnhancerScoreLOF = 1 if enhancerActivityPatientLOF < enhancerActivityControl; Otherwise geneEnhancerScoreLOF = 0

  • GOF scenario: geneEnhancerScoreGOF = 1 if enhancerActivityPatientGOF > enhancerActivityControl; Otherwise geneEnhancerScoreGOF = 0

geneFeatureScore: (dosageSensitivityScore and polycombScore)

  • dosageSensitivityScore: Used to quantify the dosage sensitivity of the candidate gene

  • polycombScore: Used to quantify the breadth of polycomb domains in gene promoters

  • geneFeatureScoreLOF = (polycombScore + dosageSensitivityScoreLOF) / 2

  • geneFeatureScoregOF = (polycombScore + dosageSensitivityScoreGOF) / 2

Long-range pathomechanisms \[ PSlongRangeLOF = \frac{\text{geneEnhancerScoreLOF} + \text{geneExpressionScore} + \text{geneFeatureScoreLOF} + \text{genePhenoScore}}{4} \]

If the SV is predicted to cause ectopic interactions between the candidate gene and non-cognate enhancers (TAD fusion, neo-TAD formation, or TAD shuffling), the geneExpressionScore is not considered relevant: \[ PSlongRangeGOF1 = \frac{\text{geneEnhancerScoreGOF} + \text{geneFeatureScoreGOF} + \text{genePhenoScore}}{3} \] If the SV is not predicted to cause ectopic interactions between the candidate gene and non-cognate enhancers (intra-TAD duplication), the geneExpressionScore is considered important: \[ PSlongRangeGOF2 = \frac{\text{geneEnhancerScoreGOF} + \text{geneExpressionScore} + \text{geneFeatureScoreGOF} + \text{genePhenoScore}}{4} \]

Coding pathomechanisms

  • If the candidate gene is deleted or truncated, only LOF will be evaluated and the PS for GOF will be set to 0; If the gene is duplicated, only GOF will be assessed and the PS for LOF will be set to 0
  • If the SVs cause gene duplication/s, long-range pathomechanisms involving enhancer adoption and formation of new TADs can occur: For SVs resulting in gene duplications, coding and long-range GOF PS are computed and the highest one is selected as the most likely pathomechanism \[ PScodingLOF = \frac{\text{geneExpressionScore} + \text{geneFeatureScoreLOF} + \text{genePhenoScore}}{3} \] \[ PScodingGOF = \frac{\text{geneExpressionScore} + \text{geneFeatureScoreGOF} + \text{genePhenoScore}}{3} \]

Run POSTRE as R function

  1. Download POSTRE repository (“Code” Github button)

  2. In the R script, load POSTRE R function: source(file = “/PathTo/Postre-main/Postre_app/POSTRE_multiSV.R”)

  3. POSTRE R function, named POSTRE_multiSV(), is already loaded in the R environment

    • Two arguments: SVs (Dataframe with 7 column containing the SVs information), pathTo_Postre_app_folder (downloaded in Step 1)
  4. Output

    • pathogenicityPrediction_per_SV: This table provides a pathogenic prediction for each of the SVs and associated phenotypes analyzed

    • geneStats_recurrencyAndPathomech: This table contains an aggregation of the pathogenic predictions per gene, phenotype and pathogenic mechanism (coding, long-range)

    • errors: Contains the ids of the SVs which rised an error (if they occur) during their interpretation.

Reproducing results from Table 1

# Loading POSTRE R function
source(file = "C:/Users/momom/Desktop/Independent Project/Postre-main/Postre_app/POSTRE_multiSV.R")

# Loading SVs patients 
svs_patients <- read.delim(file = "C:/Users/momom/Desktop/Independent Project/Postre-main/testFiles/Table1_LongRange_SVs.tsv",
                           sep = "\t",
                           stringsAsFactors = FALSE,
                           header = FALSE)

# Running POSTRE R function, with default parameters
res_svs <- POSTRE_multiSV(SVs = svs_patients, pathTo_Postre_app_Folder = "C:/Users/momom/Desktop/Independent Project/Postre-main/Postre_app/")
knitr::kable(res_svs$pathogenicityPrediction_per_SV)
SVid Phenotype PathogenicScore Pathogenic Ncausative Causative genes Ncandidates Candidate genes (Pathogenic Score)
Patient_Nr1 head_neck 1.00 Yes 1 TFAP2A 6 TFAP2A(1),FBXL4(0.67),POU3F2(0.66),EEF1E1(0.61),MMS22L(0.49),SLC35B3(0.41)
Patient_Nr2 head_neck 1.00 Yes 1 SOX9 1 SOX9(1)
Patient_Nr3 limbs 0.99 Yes 1 KCNJ2 5 KCNJ2(0.99),SOX9(0.75),MAP2K6(0.37),KCNJ16(0.33),ABCA5(0.05)
Patient_Nr4 limbs 0.98 Yes 1 PAX3 5 PAX3(0.98),EPHA4(0.5),SGPP2(0.5),STK11IP(0.38),SLC4A3(0.32)
Patient_Nr5 limbs 0.98 Yes 1 IHH 46 IHH(0.98),DNAJB2(0.67),OBSL1(0.67),TUBA4A(0.64),DES(0.57),PTPRN(0.5),RESP18(0.5),SPEG(0.5),TUBA4B(0.5),ATG9A(0.5),WNT6(0.49),STK11IP(0.38),EPHA4(0.38),WNT10A(0.38),STK36(0.36),RNF25(0.36),CHPF(0.34),ABCB6(0.33),ANKZF1(0.33),CNPPD1(0.33),DNPEP(0.33),GLB1L(0.33),NHEJ1(0.33),SLC23A3(0.33),STK16(0.33),ZFAND2B(0.33),SLC4A3(0.32),BCS1L(0.25),TTLL4(0.25),CDK5R2(0.24),CRYBA2(0.23),USP37(0.23),GMPPA(0.19),ASIC4(0.17),ZNF142(0.16),FEV(0.13),CYP27A1(0.12),PLCD4(0.02),TMEM198(0.01),CFAP65(0),CNOT9(0),INHA(0),PRKAG3(0),RETREG2(0),SPEGNB(0),VIL1(0)
Patient_Nr6 limbs 0.67 No 0 7 LMBR1(0.67),DNAJB6(0.62),UBE3C(0.44),SHH(0.44),MNX1(0.25),RNF32(0.16),NOM1(0.13)
knitr::kable(res_svs$geneStats_recurrencyAndPathomech)
Gene Phenotype N SVs N Long-Range N Coding SVid Long-Range Coding
TFAP2A_head_neck TFAP2A head_neck 1 1 0 Patient_Nr1 Patient_Nr1
SOX9_head_neck SOX9 head_neck 1 1 0 Patient_Nr2 Patient_Nr2
KCNJ2_limbs KCNJ2 limbs 1 1 0 Patient_Nr3 Patient_Nr3
PAX3_limbs PAX3 limbs 1 1 0 Patient_Nr4 Patient_Nr4
IHH_limbs IHH limbs 1 1 0 Patient_Nr5 Patient_Nr5

Reproducing results from Table 2

source(file = "C:/Users/momom/Desktop/Independent Project/Postre-main/Postre_app/POSTRE_multiSV.R")

svs_patients <- read.delim(file = "C:/Users/momom/Desktop/Independent Project/Postre-main/testFiles/Table2_LongRange_SVs.tsv",
                           sep = "\t",
                           stringsAsFactors = FALSE,
                           header = FALSE)
res_svs <- POSTRE_multiSV(SVs = svs_patients, pathTo_Postre_app_Folder = "C:/Users/momom/Desktop/Independent Project/Postre-main/Postre_app/")
knitr::kable(res_svs$pathogenicityPrediction_per_SV)
SVid Phenotype PathogenicScore Pathogenic Ncausative Causative genes Ncandidates Candidate genes (Pathogenic Score)
SATB2_Nr1 neurodevelopmental 1.00 Yes 1 SATB2 12 SATB2(1),CSTF3(0.75),HIPK3(0.63),CD59(0.63),FBXO3(0.61),QSER1(0.55),PRRG4(0.5),TCP11L1(0.35),C11orf91(0.33),DEPDC7(0.33),FTCDNL1(0),KIAA1549L(0)
SATB2_Nr2 neurodevelopmental 1.00 Yes 3 SATB2,SOX2,FXR1 6 SATB2(1),SOX2(1),FXR1(0.88),DNAJC19(0.67),ATP11B(0.43),FTCDNL1(0)
SATB2_Nr3 neurodevelopmental 1.00 Yes 2 SATB2,DMRTA2 7 SATB2(1),DMRTA2(0.83),CDKN2C(0.67),FAF1(0.67),C1orf185(0.33),RNF11(0.33),FTCDNL1(0)
SATB2_Nr4 neurodevelopmental 1.00 Yes 1 SATB2 6 SATB2(1),ASAH1(0.75),PSD3(0.64),NAT1(0.33),NAT2(0.33),FTCDNL1(0)
SATB2_Nr5 neurodevelopmental 1.00 Yes 1 SATB2 12 SATB2(1),GRAMD1B(0.75),CLMP(0.63),HSPA8(0.63),BSX(0.62),SORL1(0.62),UBASH3B(0.43),BLID(0.25),CRTAM(0.25),FTCDNL1(0),JHY(0),SC5D(0)
SATB2_Nr6 neurodevelopmental 1.00 Yes 1 SATB2 10 SATB2(1),NCK2(0.5),TGFBRAP1(0.44),C2orf49(0.4),RGPD3(0.33),UXS1(0.33),FHL2(0.01),CD8B2(0),ECRG4(0),FTCDNL1(0)
ARX_Nr1 neurodevelopmental 0.89 Yes 3 ARX,EIF2S3,KLHL15 13 ARX(0.89),EIF2S3(0.87),KLHL15(0.85),PCYT1B(0.62),POLA1(0.59),APOO(0.5),SAT1(0.5),ZFX(0.43),PDK3(0.36),ACOT9(0.3),CXorf58(0.25),SUPT20HL1(0),SUPT20HL2(0)
FOXG1_Nr1 neurodevelopmental 1.00 Yes 1 FOXG1 8 FOXG1(1),G2E3(0.56),PRKD1(0.53),SCFD1(0.53),STAC(0.5),ARPP21(0.33),DCLK3(0.33),TRANK1(0.2)
FOXG1_Nr2 neurodevelopmental 1.00 Yes 1 FOXG1 8 FOXG1(1),STMN2(0.74),G2E3(0.56),IL7(0.55),PRKD1(0.53),SCFD1(0.53),PKIA(0.5),ZC2HC1A(0)
FOXG1_Nr3 neurodevelopmental 1.00 Yes 1 FOXG1 5 FOXG1(1),G2E3(0.56),PRKD1(0.53),SCFD1(0.53),ADGRL2(0)
FOXG1_Nr4 neurodevelopmental 1.00 Yes 1 FOXG1 5 FOXG1(1),EPHA5(0.63),G2E3(0.56),PRKD1(0.53),SCFD1(0.53)
FOXG1_Nr5 neurodevelopmental 1.00 Yes 1 FOXG1 9 FOXG1(1),NFIL3(0.74),AUH(0.67),ROR2(0.67),G2E3(0.56),PRKD1(0.53),SCFD1(0.53),SPTLC1(0.4),PRSS47(0)
FOXG1_Nr6 neurodevelopmental 1.00 Yes 1 FOXG1 6 FOXG1(1),G2E3(0.56),PRKD1(0.53),SCFD1(0.53),CAMK2D(0.48),ARSJ(0.17)
FOXG1_Nr7 neurodevelopmental 1.00 Yes 1 FOXG1 5 FOXG1(1),G2E3(0.56),PRKD1(0.53),SCFD1(0.53),WDFY1(0.35)
FOXG1_Nr8 neurodevelopmental 1.00 Yes 1 FOXG1 4 FOXG1(1),G2E3(0.56),SCFD1(0.53),PRKD1(0.38)
FOXG1_Nr9 neurodevelopmental 1.00 Yes 1 FOXG1 4 FOXG1(1),G2E3(0.56),SCFD1(0.53),PRKD1(0.38)
FOXG1_Nr10 neurodevelopmental 1.00 Yes 1 FOXG1 4 FOXG1(1),G2E3(0.56),SCFD1(0.53),PRKD1(0.38)
FOXG1_Nr11 neurodevelopmental 1.00 Yes 1 FOXG1 4 FOXG1(1),G2E3(0.56),SCFD1(0.53),PRKD1(0.38)
MEF2C_Nr1 neurodevelopmental 0.88 Yes 1 MEF2C 15 MEF2C(0.88),MCOLN3(0.67),DDAH1(0.63),CETN3(0.55),COL24A1(0.5),C1orf52(0.5),LYSMD3(0.33),BCL10(0.33),SYDE2(0.33),MBLAC2(0.3),POLR3G(0.25),ZNHIT6(0.07),ADGRV1(0),CCN1(0),DNAI3(0)
MEF2C_Nr2 neurodevelopmental 0.88 Yes 1 MEF2C 16 MEF2C(0.88),COL1A2(0.67),CETN3(0.55),PON3(0.5),PPP1R9A(0.5),SGCE(0.5),PEG10(0.48),LYSMD3(0.33),ASB4(0.33),CASD1(0.33),PDK4(0.33),PON1(0.33),PON2(0.33),MBLAC2(0.3),POLR3G(0.25),ADGRV1(0)
MEF2C_Nr3 neurodevelopmental 0.88 Yes 1 MEF2C 7 MEF2C(0.88),CETN3(0.55),CDH10(0.5),LYSMD3(0.33),MBLAC2(0.33),POLR3G(0.33),ADGRV1(0)
MEF2C_Nr4 neurodevelopmental 0.88 Yes 1 MEF2C 27 MEF2C(0.88),ZYG11B(0.75),LRP8(0.75),ORC1(0.67),SCP2(0.67),PRPF38A(0.63),ZFYVE9(0.62),CETN3(0.55),BTF3L4(0.5),CC2D1B(0.5),ECHDC2(0.5),PODN(0.5),TXNDC12(0.5),MAGOH(0.49),GPX7(0.49),KTI12(0.37),LYSMD3(0.33),CPT2(0.33),SLC1A7(0.33),ZYG11A(0.33),MBLAC2(0.3),POLR3G(0.25),ADGRV1(0),COA7(0),CZIB(0),SHISAL2A(0),TUT4(0)
MEF2C_Nr5 neurodevelopmental 0.88 Yes 2 MEF2C,CHD2 10 MEF2C(0.88),CHD2(0.83),CETN3(0.55),RGMA(0.5),LYSMD3(0.33),MBLAC2(0.33),POLR3G(0.33),MCTP2(0.33),ADGRV1(0),LOC105370980(0)
MEF2C_Nr6 neurodevelopmental 0.88 Yes 1 MEF2C 6 MEF2C(0.88),CETN3(0.4),LYSMD3(0.33),POLR3G(0.25),MBLAC2(0.07),ADGRV1(0)
MEF2C_Nr7 neurodevelopmental 0.88 Yes 1 MEF2C 6 MEF2C(0.88),CETN3(0.55),LYSMD3(0.33),MBLAC2(0.3),POLR3G(0.25),ADGRV1(0)
MEF2C_Nr8 neurodevelopmental 0.88 Yes 1 MEF2C 6 MEF2C(0.88),CETN3(0.4),LYSMD3(0.11),MBLAC2(0.07),ADGRV1(0),POLR3G(0)
MEF2C_Nr9 neurodevelopmental 0.88 Yes 1 MEF2C 7 MEF2C(0.88),ARRDC3(0.75),CETN3(0.4),LYSMD3(0.11),MBLAC2(0.07),ADGRV1(0),POLR3G(0)
MEF2C_Nr10 neurodevelopmental 1.00 Yes 2 NR2F1,MEF2C 8 NR2F1(1),MEF2C(0.88),ARRDC3(0.66),CETN3(0.4),LYSMD3(0.11),MBLAC2(0.07),ADGRV1(0),POLR3G(0)
MEF2C_Nr11 neurodevelopmental 1.00 Yes 2 NR2F1,MEF2C 12 NR2F1(1),MEF2C(0.88),FAM172A(0.67),ARRDC3(0.66),CETN3(0.4),KIAA0825(0.25),LYSMD3(0.11),MBLAC2(0.07),ADGRV1(0),POLR3G(0),POU5F2(0),SLF1(0)
BCL11B_Nr1 neurodevelopmental 1.00 Yes 1 BCL11B 3 BCL11B(1),VRK1(0.75),PPARGC1A(0.5)
BCL11B_Nr2 neurodevelopmental 1.00 Yes 1 BCL11B 6 BCL11B(1),VRK1(0.75),EDNRA(0.67),ARHGAP10(0.5),TMEM184C(0.43),PRMT9(0)
SLC2A1_Nr1 neurodevelopmental 0.88 Yes 2 SLC2A1,MCOLN1 34 SLC2A1(0.88),MCOLN1(0.81),ELAVL1(0.67),PNPLA6(0.67),MAP2K7(0.67),CAMSAP3(0.65),CCL25(0.5),CLEC4G(0.5),XAB2(0.5),STXBP2(0.5),EVI5L(0.49),TIMM44(0.48),ARHGEF18(0.47),ZNF358(0.36),CD209(0.33),CERS4(0.33),CLEC4M(0.33),CTXN1(0.33),ERMAP(0.33),FBN3(0.33),FCER2(0.33),LRRC8E(0.33),PCP2(0.33),PEX11G(0.33),RETN(0.33),SNAPC2(0.33),TRAPPC5(0.33),ZNF691(0.33),MCEMP1(0),PET100(0),PRR36(0),SVBP(0),TEX45(0),TGFBR3L(0)
NR2F2_Nr1 neurodevelopmental 0.50 No 0 1 NR2F2(0.5)
CTNNA2_Nr1 neurodevelopmental 0.38 No 0 3 GLRA3(0.38),ADAM29(0.25),HPGD(0.25)
DLX5&6_Nr1 limbs 0.99 Yes 1 DLX5 5 DLX5(0.99),DLX6(0.67),SLC25A13(0.33),DYNC1I1(0.11),SEM1(0)
DLX5&6_Nr2 limbs 0.99 Yes 1 DLX5 5 DLX5(0.99),DLX6(0.67),SLC25A13(0.33),DYNC1I1(0.11),SEM1(0)
DLX5&6_Nr3 limbs 0.99 Yes 1 DLX5 5 DLX5(0.99),DLX6(0.67),SLC25A13(0.33),DYNC1I1(0.11),SEM1(0)
DLX5&6_Nr4 limbs 0.99 Yes 1 DLX5 5 DLX5(0.99),DLX6(0.67),SLC25A13(0.33),DYNC1I1(0.11),SEM1(0)
DLX5&6_Nr5 limbs 0.99 Yes 1 DLX5 5 DLX5(0.99),DLX6(0.67),SLC25A13(0.33),DYNC1I1(0.11),SEM1(0)
DLX5&6_Nr6 limbs 0.99 Yes 1 DLX5 13 DLX5(0.99),DLX6(0.67),FKBP14(0.66),CHN2(0.55),PLEKHA8(0.5),SCRN1(0.5),TRIL(0.5),PRR15(0.49),CPVL(0.45),WIPF3(0.38),DYNC1I1(0.33),SLC25A13(0.33),SEM1(0)
DLX5&6_Nr7 limbs 0.99 Yes 2 DLX5,ODC1 9 DLX5(0.99),ODC1(0.83),DLX6(0.67),RRM2(0.37),HPCAL1(0.33),NOL10(0.28),SLC25A13(0.25),DYNC1I1(0.08),SEM1(0)