Model is empty!

[Kai6662]

When I tried to use my own data, I got the bug information. I tested using the sample data, it worked. But my data was failed to run.

fit <- bseqsc_proportions(bulk, B, verbose = TRUE)

• Data features: 'TSPAN6', 'TNMD', ..., 'AC022726.2' (53,145 total)
• Basis features: 'GZMA', 'NKG7', ..., 'MYH11' (37 total)
• Common features: 'CD79B', 'CD22', ..., 'LTB' (37 total)
• Converting to linear scale
• Writing input files ... OK
• Running CIBERSORT ...
  Show Traceback

Rerun with Debug
Error in predict.svm(ret, xhold, decision.values = TRUE) :
Model is empty!

Traceback:
12.
stop("Model is empty!")
11.
predict.svm(ret, xhold, decision.values = TRUE)
10.
predict(ret, xhold, decision.values = TRUE)
9.
na.action(predict(ret, xhold, decision.values = TRUE))
8.
svm.default(X, y, type = "nu-regression", kernel = "linear",
nu = nus, scale = F)
7.
svm(X, y, type = "nu-regression", kernel = "linear", nu = nus,
scale = F) at CIBERSORT.R#58
6.
FUN(X[[i]], ...)
5.
lapply(X, FUN, ...)
4.
mclapply(1:svn_itor, res, mc.cores = 1) at CIBERSORT.R#62
3.
CoreAlg(X, y, absolute, abs_method) at CIBERSORT.R#200
2.
CIBERSORT(xf, yf, ...)
1.
bseqsc_proportions(bulk, B, verbose = TRUE)

[josemss]

When I tried your tutorial, with your own data, I get the error:

fit <- bseqsc_proportions(eset, B, verbose = TRUE)
Error in predict.svm(ret, xhold, decision.values = TRUE) :
Model is empty!

Traceback:

16: stop("Model is empty!")
15: predict.svm(ret, xhold, decision.values = TRUE)
14: predict(ret, xhold, decision.values = TRUE)
13: na.action(predict(ret, xhold, decision.values = TRUE))
12: svm.default(X, y, type = "nu-regression", kernel = "linear",
nu = nus, scale = F)
11: svm(X, y, type = "nu-regression", kernel = "linear", nu = nus,
scale = F) at CIBERSORT.R#103
10: FUN(X[[i]], ...)
9: lapply(1:svn_itor, res) at CIBERSORT.R#109
8: CoreAlg(X, y, absolute, abs_method) at CIBERSORT.R#247
7: CIBERSORT(sig_matrix = x, mixture_file = y) at CIBERSORT.R#289
6: eval(ei, envir)
5: eval(ei, envir)
4: withVisible(eval(ei, envir))
3: source(cib, local = env)
2: bseqsc_config(error = TRUE)
1: bseqsc_proportions(eset, B, verbose = TRUE)

Thanks.

[BirongZhang]

Hi all,

I got the same error:

> results <-cibersort(sig_matrix, mixture_file, perm=50, QN=TRUE)
Error in predict.svm(ret, xhold, decision.values = TRUE) : 
  Model is empty!

> traceback()
12: stop(condition)
11: signalConditions(obj, exclude = getOption("future.relay.immediate", 
        "immediateCondition"), resignal = resignal, ...)
10: signalConditionsASAP(obj, resignal = FALSE, pos = ii)
9: resolve.list(y, result = TRUE, stdout = stdout, signal = signal, 
       force = TRUE)
8: resolve(y, result = TRUE, stdout = stdout, signal = signal, force = TRUE)
7: value.list(futures)
6: future::value(futures)
5: furrr_template(args = x, fn = fn, dots = dots, n = n, options = options, 
       progress = progress, type = type, map_fn = map_fn, names = names, 
       env_globals = env_globals, expr = expr, extract = furrr_map_extract)
4: furrr_map_template(x = .x, fn = .f, dots = list(...), options = .options, 
       progress = .progress, type = "list", map_fn = purrr::map, 
       env_globals = .env_globals)
3: future_map(1:svn_itor, res)
2: CoreAlg(X, y)
1: cibersort(sig_matrix, mixture_file, perm = 50, QN = TRUE)

here is LM22.txt:

Here is my mixture file:

Any advice would be appreciated, thanks!

[zpingfeng]

#11 (comment) I got similar error when I run cibersort with some of my own data. Now I figured our the problem is that the labels of geneID are slightly different. It works after I fixed the labels