|
a |
|
b/weighted.stats.R |
|
|
1 |
weighted.stats = |
|
|
2 |
function (x, w, conditions) |
|
|
3 |
{ |
|
|
4 |
n = ncol(x) #number of samples |
|
|
5 |
p = nrow(x) #number of genes |
|
|
6 |
nclass = length(unique(conditions)) #number of class |
|
|
7 |
if(is.factor(conditions)) {cNames = sort(levels(conditions))} |
|
|
8 |
if (is.numeric(conditions)) {cNames = as.character(sort(unique(conditions)))} |
|
|
9 |
|
|
|
10 |
WM = WS = wSum = se.scale = matrix(0, p, nclass) |
|
|
11 |
rownames(WS) = rownames(WM) = rownames(wSum) = rownames(se.scale) = rownames(x) |
|
|
12 |
colnames(WS) = colnames(WM) = colnames(wSum) = colnames(se.scale) = cNames |
|
|
13 |
|
|
|
14 |
c.ind = as.numeric(conditions) |
|
|
15 |
|
|
|
16 |
w.mean00 = |
|
|
17 |
function (x, w) |
|
|
18 |
{ |
|
|
19 |
wm = NULL |
|
|
20 |
|
|
|
21 |
for (i in 1:p) |
|
|
22 |
{ |
|
|
23 |
wm0 = sum(w[i,]*x[i,]) / sum(w[i,]) |
|
|
24 |
wm = c(wm, wm0) |
|
|
25 |
} |
|
|
26 |
return(wm) |
|
|
27 |
} |
|
|
28 |
|
|
|
29 |
w.mean = |
|
|
30 |
function (x, w, conditions) |
|
|
31 |
{ |
|
|
32 |
for (j in 1:nclass) |
|
|
33 |
{ |
|
|
34 |
WM[,j] = w.mean00(x[,c.ind == j], w[,c.ind == j]) |
|
|
35 |
} |
|
|
36 |
return(WM) |
|
|
37 |
} |
|
|
38 |
|
|
|
39 |
w.sd = |
|
|
40 |
function (x, w, conditions) |
|
|
41 |
{ |
|
|
42 |
w.sd00 = |
|
|
43 |
function (x, w) |
|
|
44 |
{ |
|
|
45 |
ws = NULL |
|
|
46 |
|
|
|
47 |
w.sd0 = |
|
|
48 |
function (x, w) |
|
|
49 |
{ |
|
|
50 |
sumw = sum(w) |
|
|
51 |
sumw.sq = sum(w)^2 |
|
|
52 |
w.sq = sum(w^2) |
|
|
53 |
denom = sum(w * ((x - mean(x))^2)) |
|
|
54 |
sqrt((sumw * denom) / (sumw.sq - w.sq)) |
|
|
55 |
} |
|
|
56 |
|
|
|
57 |
for (i in 1:p) |
|
|
58 |
{ |
|
|
59 |
ws0 = w.sd0(x[i,], w[i,]) |
|
|
60 |
ws = c(ws, ws0) |
|
|
61 |
} |
|
|
62 |
|
|
|
63 |
return(ws) |
|
|
64 |
} |
|
|
65 |
|
|
|
66 |
for (j in 1:nclass) |
|
|
67 |
{ |
|
|
68 |
WS[,j] = w.sd00(x[,c.ind == j], w[,c.ind == j]) |
|
|
69 |
} |
|
|
70 |
return(WS) |
|
|
71 |
} |
|
|
72 |
|
|
|
73 |
weightedMean = w.mean00(x, w) #Overall weighted mean |
|
|
74 |
weightedMean.C = w.mean(x, w, conditions) #Weighted means for each group |
|
|
75 |
weightedSD.C = w.sd(x, w, conditions) #Weighted standard deviations for each group |
|
|
76 |
#weightedSD.pooled = weightedSD.C |
|
|
77 |
|
|
|
78 |
|
|
|
79 |
for (i in 1:nclass) |
|
|
80 |
{ |
|
|
81 |
tmp = w[,which(c.ind == i)] |
|
|
82 |
rSum.tmp = rowSums(tmp) |
|
|
83 |
|
|
|
84 |
wSum[,i] = rSum.tmp |
|
|
85 |
} |
|
|
86 |
|
|
|
87 |
weightedSD.pooled = sqrt(rowSums((wSum-1) * (weightedSD.C^2)) / (rowSums(wSum) - nclass)) |
|
|
88 |
se.scale = sqrt(1 / wSum + 1 / rowSums(wSum)) |
|
|
89 |
|
|
|
90 |
s0 = median(weightedSD.pooled) |
|
|
91 |
|
|
|
92 |
delta = (weightedMean.C - weightedMean)/(se.scale*(weightedSD.pooled + s0)) |
|
|
93 |
|
|
|
94 |
|
|
|
95 |
weightedSD.pooled = sqrt(rowSums(as.data.frame(weightedSD.pooled)) / (n - nclass)) |
|
|
96 |
stats = list(n = n, p = p, nclass = nclass, se.scale = se.scale, weightedMean = weightedMean, weightedMean.C = weightedMean.C, weightedSD.C = weightedSD.C, weightedSD.pooled = weightedSD.pooled, delta = delta) |
|
|
97 |
|
|
|
98 |
return(stats) |
|
|
99 |
} |