survival analysis with several conditions
0
0
Entering edit mode
2.2 years ago
Learner ▴ 250

Hello guys,

I have several conditions a WT and two drug treated patients conditions. I want to calculate the significance differences between drug_1 (treated) versus WT and drug_2 (treated) versus WT.

  Sample Hour Condition
1       WT    0         1
2       WT    0         1
3       WT    0         1
4       WT    0         1
5       WT    0         1
6       WT    0         1
7       WT    0         1
8       WT    0         1
9       WT    0         1
10      WT    0         1
11  Drug_1    0         1
12  Drug_1    0         1
13  Drug_1    0         1
14  Drug_1    0         1
15  Drug_1    0         1
16  Drug_1    0         1
17  Drug_1    0         1
18  Drug_1    0         1
19  Drug_1    0         1
20  Drug_1    0         1
21  Drug_1    0         1
22  Drug_1    0         1
23  Drug_2    0         1
24  Drug_2    0         1
25  Drug_2    0         1
26  Drug_2    0         1
27  Drug_2    0         1
28  Drug_2    0         1
29  Drug_2    0         1
30  Drug_2    0         1
31      WT    1         1
32      WT    1         1
33      WT    1         1
34      WT    1         1
35      WT    1         1
36      WT    1         1
37      WT    1         0
38      WT    1         1
39      WT    1         1
40      WT    1         1
41  Drug_1    1         1
42  Drug_1    1         1
43  Drug_1    1         1
44  Drug_1    1         1
45  Drug_1    1         1
46  Drug_1    1         1
47  Drug_1    1         1
48  Drug_1    1         1
49  Drug_1    1         1
50  Drug_1    1         1
51  Drug_1    1         1
52  Drug_1    1         1
53  Drug_2    1         1
54  Drug_2    1         1
55  Drug_2    1         1
56  Drug_2    1         1
57  Drug_2    1         1
58  Drug_2    1         1
59  Drug_2    1         1
60  Drug_2    1         0
61      WT    5         1
62      WT    5         1
63      WT    5         0
64      WT    5         0
65      WT    5         1
66      WT    5         1
67      WT    5         1
68      WT    5         1
69      WT    5         1
70  Drug_1    5         1
71  Drug_1    5         1
72  Drug_1    5         1
73  Drug_1    5         1
74  Drug_1    5         1
75  Drug_1    5         0
76  Drug_1    5         1
77  Drug_1    5         1
78  Drug_1    5         1
79  Drug_1    5         1
80  Drug_1    5         1
81  Drug_1    5         1
82  Drug_2    5         1
83  Drug_2    5         1
84  Drug_2    5         1
85  Drug_2    5         1
86  Drug_2    5         1
87  Drug_2    5         0
88  Drug_2    5         0
89      WT   10         1
90      WT   10         0
91      WT   10         0
92      WT   10         0
93      WT   10         1
94      WT   10         1
95      WT   10         1
96  Drug_1   10         1
97  Drug_1   10         1
98  Drug_1   10         1
99  Drug_1   10         1
100 Drug_1   10         0
101 Drug_1   10         1
102 Drug_1   10         1
103 Drug_1   10         1
104 Drug_1   10         1
105 Drug_1   10         1
106 Drug_1   10         1
107 Drug_1   10         1
108 Drug_2   10         1
109 Drug_2   10         1
110 Drug_2   10         0
111 Drug_2   10         1
112     WT   11         0
113     WT   11         1
114     WT   11         1
115     WT   11         1
116 Drug_1   11         1
117 Drug_1   11         0
118 Drug_1   11         1
119 Drug_1   11         1
120 Drug_1   11         1
121 Drug_1   11         1
122 Drug_1   11         1
123 Drug_1   11         1
124 Drug_1   11         1
125 Drug_1   11         1
126 Drug_2   11         1
127 Drug_2   11         1
128 Drug_2   11         1
129 Drug_2   11         1


Now, I have few questions

1. how can I perform cox proportional hazard analysis on this data?

2. do I need any multiway correction test?

Thanks

R • 613 views
0
Entering edit mode
1. To my understanding, drug_1, drug_2 and WT are independent groups/labels. Assure that your wanting to compare overall survival of patients after taking orally drug_1 versus WT, and those after taking orally drug_2 versus WT, so the question is that how many patients are in group 1 (drug_1), group 2 (drug_2), and group 3 (WT)? Basically, your current data are insuffienct to run survival analysis.

2. I don't get you. You mean should you compute q-value for survival analysis?

0
Entering edit mode

@huynguyen96.dnu Hi, Thanks for reacting on my post, why do you think it is insufficient? of course the real data is way bigger (I cannot post the whole thing because it is huge) so if you show me the code, I will try to adjust it on a bigger data

2- should I do drug_1 and drug2 interaction or Tukey test etc?

0
Entering edit mode

1-Assure that you divided the above data frame into two sub data frames: one (I call it as 'df' in R) whose 'Sample' column only includes group1 (drug_1) versus group3 (WT), the other (I call it as 'df1' in R) whose 'Sample' column only includes group2 (drug_2) versus group3 (WT) Then, for the first sub data frame 'df' (do the same for 'df1'), the first code lines may be something like:

library(survival);
summary(coxph(
Surv(Hour, Condition) ~ as.factor(groups),
data = df,
ties = "exact"
))


in which, 'groups' which is integer type in R includes how many patients are in group 1 (drug_1) and group 3 (WT). It looks like:

Patient 1 Patient2 Patient3 ..... Patient n

-----1 --------- 3--------- 1------ ..... --- 1

2-I don't think it is important to do that