[sthor]

Passenger versus Driver

Hi,

how do you interprete the oncofuse output based on the definitions below?

low P_VAL_CORR => likely passenger?
high DRIVER_PROB => likely driver?
EXPRESSION_GAIN => how are 5'/3' expressions computed?

I have these two lines for instance which seem contradictory:
1130 0 KDR Yes Exon 6 1 1138 2 2 5.616331914947E-5 0.9999999839119682 0.7550226777191273 Pleckstrin homology-like domain[Domain];Arf GTPase activating protein[Family];Pleckstrin homology domain[Domain];Arfaptin homology (AH) domain/BAR domain[Domain];Src homology-3 domain[Domain];Ankyrin repeat-containing domain[Domain];Ankyrin repeat[Repeat] Immunoglobulin-like domain[Domain];Immunoglobulin subtype 2[Domain];Immunoglobulin subtype[Domain];Serine-threonine/tyrosine-protein kinase catalytic domain[Domain];Immunoglobulin-like fold[Domain];Protein kinase, ATP binding site[Binding_site];Tyrosine-protein kinase, receptor class III, conserved site[Conserved_site];Immunoglobulin I-set[Domain];Tyrosine-protein kinase, active site[Active_site];Protein kinase-like domain[Domain];Immunoglobulin V-set domain[Domain];Tyrosine-protein kinase, catalytic domain[Domain];Protein kinase domain[Domain];Vascular endothelial growth factor receptor 2 (VEGFR2)[Family] GRB2;SRC KDR;NCK1;PLCG1;PTPN6;SHC1;SRC;VEGFA;VEGFC;NRP1;SH2D2A;SHC2;FLT4;FYN;GRB2;GRB10 0.031693123551895894 0.060912399971582896 0.0011108580362268734 0.08982267991811016 0.03377877733466933 0.01758896057545965
679 0 CLTC Yes Exon 10 31 1159 0 0 4.159324576929766E-4 0.9999996425673524 -0.33165129627027434 Zinc finger, C2H2-like[Domain] Clathrin, heavy chain/VPS, 7-fold repeat[Repeat];Clathrin, heavy chain[Family];Tetratricopeptide-like helical[Domain];Armadillo-type fold[Domain] Zinc finger, C2H2-like[Domain] CLTC;CLTA;CLTB 0.00561979753630853 0.0 3.7249030480169165E-5 0.0 0.0011034671362001604 0.018050986438399443


P_VAL_CORR

The Bayesian probability of fusion being a passenger (class 0), given as Bonferroni-corrected P-value.
DRIVER_PROB

The Bayesian probability of fusion being a driver (class 1).
EXPRESSION_GAIN

Expression gain of fusion calculated as:

[(expression of 5' gene)/(expression of 3' gene)]-1

Thanks,

[mikesh]

Hi,

how do you interprete the oncofuse output based on the definitions below?

low P_VAL_CORR => likely passenger?
high DRIVER_PROB => likely driver?
EXPRESSION_GAIN => how are 5'/3' expressions computed?

I have these two lines for instance which seem contradictory:
1130 0 KDR Yes Exon 6 1 1138 2 2 5.616331914947E-5 0.9999999839119682 0.7550226777191273 Pleckstrin homology-like domain[Domain];Arf GTPase activating protein[Family];Pleckstrin homology domain[Domain];Arfaptin homology (AH) domain/BAR domain[Domain];Src homology-3 domain[Domain];Ankyrin repeat-containing domain[Domain];Ankyrin repeat[Repeat] Immunoglobulin-like domain[Domain];Immunoglobulin subtype 2[Domain];Immunoglobulin subtype[Domain];Serine-threonine/tyrosine-protein kinase catalytic domain[Domain];Immunoglobulin-like fold[Domain];Protein kinase, ATP binding site[Binding_site];Tyrosine-protein kinase, receptor class III, conserved site[Conserved_site];Immunoglobulin I-set[Domain];Tyrosine-protein kinase, active site[Active_site];Protein kinase-like domain[Domain];Immunoglobulin V-set domain[Domain];Tyrosine-protein kinase, catalytic domain[Domain];Protein kinase domain[Domain];Vascular endothelial growth factor receptor 2 (VEGFR2)[Family] GRB2;SRC KDR;NCK1;PLCG1;PTPN6;SHC1;SRC;VEGFA;VEGFC;NRP1;SH2D2A;SHC2;FLT4;FYN;GRB2;GRB10 0.031693123551895894 0.060912399971582896 0.0011108580362268734 0.08982267991811016 0.03377877733466933 0.01758896057545965
679 0 CLTC Yes Exon 10 31 1159 0 0 4.159324576929766E-4 0.9999996425673524 -0.33165129627027434 Zinc finger, C2H2-like[Domain] Clathrin, heavy chain/VPS, 7-fold repeat[Repeat];Clathrin, heavy chain[Family];Tetratricopeptide-like helical[Domain];Armadillo-type fold[Domain] Zinc finger, C2H2-like[Domain] CLTC;CLTA;CLTB 0.00561979753630853 0.0 3.7249030480169165E-5 0.0 0.0011034671362001604 0.018050986438399443


P_VAL_CORR

The Bayesian probability of fusion being a passenger (class 0), given as Bonferroni-corrected P-value.
DRIVER_PROB

The Bayesian probability of fusion being a driver (class 1).
EXPRESSION_GAIN

Expression gain of fusion calculated as:

[(expression of 5' gene)/(expression of 3' gene)]-1

Thanks,

Hello!

DRIVER_PROB means the probability (in Bayesian sense) of being a driver mutation.

The same in represented by P-value, the probability of being a passanger, with one major difference. P_VAL_CORR = FDR(1-DRIVER_PROB), where FDR represents Benjamini-Hochberg corrected P-values.

Expression gain is calculated based on gene expression values in normal tissue, according to library option. The corresponding routine uses the gene expression data provided with Oncofuse bundle.

So both of them are putative drivers, with promoter swapping being possible mechanism in the first case.

---

PS
Just in case, I'm adding the link to source code repository of Oncofuse containing the latest release: https://github.com/mikessh/oncofuse