R-loop STAD analysis

Hp gene detection and R-loop (lax filtering)

Summary table split by Hp detection with lax filtering, only one technology (RNAseq, WXS or WGS) need to have at least one read aligned to the Hp genome. A simple test for R-loop enrichment in Hp infected tumors would be to compare the contingency tables between having Hp and having an R-loop mutation. In the R_loop_mutation line the p-value for the chi-squared test is 0.7, there is no enrichment. I tried to filter the mutations, where I only chose SNVs with a predicted deleterious or damaging effect, since mutations could occur on the gene where the effect is benign. This reduced the number of R-loop mutations from 69 to 48, but again the p-value isn’t so low p = 0.5.

## There was an error in 'add_p()' for variable 'TNM.Stage' and test 'fisher.test', p-value omitted:
## Error in stats::fisher.test(data[[variable]], as.factor(data[[by]])): FEXACT error 7(location). LDSTP=18600 is too small for this problem,
##   (pastp=85.2035, ipn_0:=ipoin[itp=416]=158, stp[ipn_0]=84.542).
## Increase workspace or consider using 'simulate.p.value=TRUE'

Characteristic¹	N	FALSE, N = 175	TRUE, N = 92	p-value²
Hp_total_str	267	0 (0%)	36 (39%)	<0.001
R_loop_mutation	267	47 (27%)	22 (24%)	0.7
R_loop_mut_damaging	267	29 (17%)	19 (21%)	0.5
Subtype	267			>0.9
CIN		88 (50%)	48 (52%)
EBV		18 (10%)	7 (7.6%)
GS		33 (19%)	17 (18%)
MSI		36 (21%)	20 (22%)
TP53.mutation	267			0.8
0		88 (50%)	49 (53%)
1		85 (49%)	43 (47%)
NA		2 (1.1%)	0 (0%)
PIK3CA.mutation	267			0.5
0		141 (81%)	71 (77%)
1		32 (18%)	21 (23%)
NA		2 (1.1%)	0 (0%)
KRAS.mutation	267			0.2
0		161 (92%)	80 (87%)
1		12 (6.9%)	12 (13%)
NA		2 (1.1%)	0 (0%)
MSI.status	267			>0.9
MSI-H		36 (21%)	20 (22%)
MSI-L		28 (16%)	15 (16%)
MSS		111 (63%)	57 (62%)
Hypermutated	267			0.7
0		136 (78%)	74 (80%)
1		37 (21%)	18 (20%)
NA		2 (1.1%)	0 (0%)
TNM.Stage	267
Stage_IA		4 (2.3%)	4 (4.3%)
Stage_IB		11 (6.3%)	8 (8.7%)
Stage_IIA		34 (19%)	19 (21%)
Stage_IIB		38 (22%)	17 (18%)
Stage_IIIA		22 (13%)	13 (14%)
Stage_IIIB		37 (21%)	14 (15%)
Stage_IIIC		10 (5.7%)	3 (3.3%)
Stage_IV		12 (6.9%)	6 (6.5%)
X		7 (4.0%)	8 (8.7%)
Lauren.Class	267			0.3
Diffuse		42 (24%)	19 (21%)
Intestinal		112 (64%)	66 (72%)
Mixed		15 (8.6%)	3 (3.3%)
NA		6 (3.4%)	4 (4.3%)
¹ Statistics presented: n (%) ² Statistical tests performed: chi-square test of independence; Fisher's exact test

Hp gene detection and R-loop (stringent filtering)

Summary table split by Hp detection with stringent filtering, at least two technologies (RNAseq, WXS or WGS) need to have at least one read aligned to the Hp genome. In the R_loop_mutation line the p-value for the chi-squared test is 0.6, there is no enrichment. For R-loop damaging mutations, the p-value also is not significant, p = 0.5.

## There was an error in 'add_p()' for variable 'TNM.Stage' and test 'fisher.test', p-value omitted:
## Error in stats::fisher.test(data[[variable]], as.factor(data[[by]])): FEXACT error 7(location). LDSTP=18600 is too small for this problem,
##   (pastp=29.7503, ipn_0:=ipoin[itp=131]=2803, stp[ipn_0]=29.2407).
## Increase workspace or consider using 'simulate.p.value=TRUE'

Characteristic¹	N	FALSE, N = 231	TRUE, N = 36	p-value²
Hp_total_lax	267	56 (24%)	36 (100%)	<0.001
R_loop_mutation	267	58 (25%)	11 (31%)	0.6
R_loop_mut_damaging	267	39 (17%)	9 (25%)	0.3
Subtype	267			0.3
CIN		121 (52%)	15 (42%)
EBV		23 (10.0%)	2 (5.6%)
GS		43 (19%)	7 (19%)
MSI		44 (19%)	12 (33%)
TP53.mutation	267			0.12
0		113 (49%)	24 (67%)
1		116 (50%)	12 (33%)
NA		2 (0.9%)	0 (0%)
PIK3CA.mutation	267			0.4
0		186 (81%)	26 (72%)
1		43 (19%)	10 (28%)
NA		2 (0.9%)	0 (0%)
KRAS.mutation	267			0.064
0		212 (92%)	29 (81%)
1		17 (7.4%)	7 (19%)
NA		2 (0.9%)	0 (0%)
MSI.status	267			0.13
MSI-H		44 (19%)	12 (33%)
MSI-L		39 (17%)	4 (11%)
MSS		148 (64%)	20 (56%)
Hypermutated	267			0.3
0		185 (80%)	25 (69%)
1		44 (19%)	11 (31%)
NA		2 (0.9%)	0 (0%)
TNM.Stage	267
Stage_IA		5 (2.2%)	3 (8.3%)
Stage_IB		17 (7.4%)	2 (5.6%)
Stage_IIA		44 (19%)	9 (25%)
Stage_IIB		49 (21%)	6 (17%)
Stage_IIIA		34 (15%)	1 (2.8%)
Stage_IIIB		47 (20%)	4 (11%)
Stage_IIIC		12 (5.2%)	1 (2.8%)
Stage_IV		16 (6.9%)	2 (5.6%)
X		7 (3.0%)	8 (22%)
Lauren.Class	267			0.4
Diffuse		53 (23%)	8 (22%)
Intestinal		154 (67%)	24 (67%)
Mixed		17 (7.4%)	1 (2.8%)
NA		7 (3.0%)	3 (8.3%)
¹ Statistics presented: n (%) ² Statistical tests performed: chi-square test of independence; Fisher's exact test

Hp status and R-loop mutation by Subtype

To check if there is a specific subtype effect on R-loop mutation and Hp infection, we can do the same chi-squared tests on Hp vs R-loop after subsetting by subtype. You can see here that again there doesn’t seem to be a significant enrichment for R-loop mutations in Hp positive tumors in a particular subtype.

## There was an error in 'add_p()' for variable 'R_loop_mut_damaging' and test 'chisq.test', p-value omitted:
## Error in stats::chisq.test(data[[variable]], as.factor(data[[by]])): 'x' and 'y' must have at least 2 levels

## There was an error in 'add_p()' for variable 'MSI.status' and test 'chisq.test', p-value omitted:
## Error in stats::chisq.test(data[[variable]], as.factor(data[[by]])): 'x' and 'y' must have at least 2 levels

## Styling functions `bold_labels()`, `bold_levels()`, `italicize_labels()`, and `italicize_levels()` need to be re-applied after `tbl_merge()`.

Characteristic	CIN				EBV				GS				MSI
Characteristic	N	FALSE, N = 88¹	TRUE, N = 48¹	p-value²	N	FALSE, N = 18¹	TRUE, N = 7¹	p-value³	N	FALSE, N = 33¹	TRUE, N = 17¹	p-value³	N	FALSE, N = 36¹	TRUE, N = 20¹	p-value⁴
Hp_total_str	136	0 (0%)	15 (31%)	<0.001	25	0 (0%)	2 (29%)	0.070	50	0 (0%)	7 (41%)	<0.001	56	0 (0%)	12 (60%)	<0.001
R_loop_mutation	136	12 (14%)	6 (12%)	>0.9	25	3 (17%)	0 (0%)	0.5	50	4 (12%)	1 (5.9%)	0.6	56	28 (78%)	15 (75%)	>0.9
R_loop_mut_damaging	136	7 (8.0%)	4 (8.3%)	>0.9	25	0 (0%)	0 (0%)		50	1 (3.0%)	1 (5.9%)	>0.9	56	21 (58%)	14 (70%)	0.6
MSI.status	136			0.6	25			0.5	50			0.7	56
MSI-L		19 (22%)	13 (27%)			3 (17%)	0 (0%)			6 (18%)	2 (12%)
MSS		69 (78%)	35 (73%)			15 (83%)	7 (100%)			27 (82%)	15 (88%)
MSI-H														36 (100%)	20 (100%)
Hypermutated	136			0.7	25			0.5	50			0.5	56			0.5
0		85 (97%)	45 (94%)			15 (83%)	7 (100%)			31 (94%)	17 (100%)			5 (14%)	5 (25%)
1		3 (3.4%)	3 (6.2%)			3 (17%)	0 (0%)							31 (86%)	15 (75%)
NA										2 (6.1%)	0 (0%)
¹ Statistics presented: n (%) ² Statistical tests performed: chi-square test of independence; Fisher's exact test ³ Statistical tests performed: Fisher's exact test ⁴ Statistical tests performed: Fisher's exact test; chi-square test of independence

Oncoprint with everything together

I added a new track where I point out the “damaging” R-loop mutations. As you can see there isn’t a specific clustering of R-loop mutations with Hp presence.

## All mutation types: Missense_Mutation, Amplification,
## Frame_Shift_InDel, Deletion, Splice_Site, Nonsense_Mutation,
## In_Frame_InDel

Analyze mutations patterns of Hp positive and negative patients

Comparing mutated genes between 92 Hp positive and 175 Hp negative. If a gene is mutated more than once it is only counted once. The Fisher test is used to assess if the proportion of a mutated gene between the two group is different. 443 genes have a p < 0.05

## Gene list complete

No significant NFkB genes

A blank cell means the gene was not mutated in either group

There are some genes that have different mutation frequences between the two groups

I plotted the genes that have a p < 0.05 in a heatmap. There could be a mutation pattern in MSI subtype patients with HP.

looking at Subtype specific mutation patterns between Hp positive and negative patients

We can do the same fisher test for within the subtype. Only MSI and CIN subtypes had significant genes. 410 in MSi and 92 in CIN

MSI table

No significant NFkB in MSI HP mutated genes

Heatmap just for MSI subtype with significant mutation frequencies between Hp positive and negative

CIN table

No significant NFkB in CIN HP mutated genes

Heatmap just for MSI subtype with significant mutation frequencies between Hp positive and negative

Interestingly for CIN the average TMB between Hp positive and negative is significant but not for MSI

Characteristic	CIN	EBV	GS	MSI
TMB	136	1.52 (0.94)	2.12 (1.63)	0.024	25	2.48 (3.25)	1.04 (0.30)	0.080	50	3.30 (13.16)	0.84 (0.77)	0.3	56	16 (7)	21 (17)	0.3
¹ Statistics presented: mean (SD) ² Statistical tests performed: t-test

Characteristic

CIN

EBV

GS

MSI

N

FALSE, N = 88¹

TRUE, N = 48¹

p-value²

N

FALSE, N = 18¹

TRUE, N = 7¹

p-value²

N

FALSE, N = 33¹

TRUE, N = 17¹

p-value²

N

FALSE, N = 36¹

TRUE, N = 20¹

p-value²

TMB

136

1.52 (0.94)

2.12 (1.63)

0.024

25

2.48 (3.25)

1.04 (0.30)

0.080

50

3.30 (13.16)

0.84 (0.77)

0.3

56

16 (7)

21 (17)

0.3

¹ Statistics presented: mean (SD)

² Statistical tests performed: t-test

R-loop STAD analysis

Phil Cheng

11 2 2020

Hp gene detection and R-loop (lax filtering)

Hp gene detection and R-loop (stringent filtering)

Hp status and R-loop mutation by Subtype

Oncoprint with everything together

Analyze mutations patterns of Hp positive and negative patients

No significant NFkB genes

There are some genes that have different mutation frequences between the two groups

looking at Subtype specific mutation patterns between Hp positive and negative patients

MSI table

No significant NFkB in MSI HP mutated genes

Heatmap just for MSI subtype with significant mutation frequencies between Hp positive and negative

CIN table

No significant NFkB in CIN HP mutated genes

Heatmap just for MSI subtype with significant mutation frequencies between Hp positive and negative

Interestingly for CIN the average TMB between Hp positive and negative is significant but not for MSI

Unfortunately, pathway analysis on HP mutation signature overall, in MSI and in CIN didn’t overlap with any signaling pathways.