Question

DRIMseq error for paired sample analysis

0

Entering edit mode

Luca • 0

@lucapiacentini-9597

Last seen 10 weeks ago

Italy

Hello, I am doing a DTU with DRIMseq and I can get the results according to the following model: design_full <- model.matrix(~condition, data=DRIMSeq::samples(d)) where "condition" is my variable of interest (factor with 4 levels) and executing in sequence d <- dmPrecision(d, design=design_full) d <- dmFit(d, design=design_full) d <- dmTest(d, coef="condition4") # condition4= one of the four levels against reference condition

However, now I would like to do the analysis for paired data (which would be more correct for this experimental design) and I have added the variable 'subject' to the model, which indicates the pairing of samples: design_full <- model.matrix(~condition+subject, data=DRIMSeq::samples(d)) In this case, however, the analysis stops and I get the following error messages:

Error: BiocParallel errors 1 remote errors, element index: 135 731 unevaluated and other errors first remote error: Error in optimHess(par = par, fn = dm_lik_regG, gr = dm_score_regG, x = x, : non-finite value supplied by optim

Do you have any suggestions for resolving these errors? Thank you in advance. Luca

DRIMSeq • 911 views

ADD COMMENT • link 20 months ago Luca • 0

0

Entering edit mode

Your code would be easier to understand if you add a small example of your dataset and if you correctly format the code

ADD REPLY • link 20 months ago Basti ▴ 780

0

Entering edit mode

The dataset is made of 24 total samples from 6 different subjects analysed for 4 different conditions. Sample_ID= sample_1 to sample_24; subjects= subject_1 to subject_6; conditions= condition_1 to condition_4

The transcripts were quantified with Salmon and the object d was obtained as follows:

d <- dmDSdata(counts=counts, samples=my_dataframe)

and then filtered as follows:

n <- 24
n.small <- 6
d <- dmFilter(d,
          min_samps_feature_expr=n.small, 
          min_feature_expr=10,
          min_samps_feature_prop=n.small, 
          min_feature_prop=0.1,
          min_samps_gene_expr=n.small, 
          min_gene_expr=10)

# > d An object of class dmDSdata  with 8444 genes and 24 samples

Then, I designed the model matrix:

design_full <- model.matrix(~condition+subject, data=DRIMSeq::samples(d))

and ran the analysis:

set.seed(1)
system.time({
d <- dmPrecision(d, design=design_full)
d <- dmFit(d, design=design_full)
d <- dmTest(d, coef="condition_4")})

From here the error as reported above. Hope this is clear enough now. Thanks!

ADD REPLY • link 20 months ago Luca • 0

0

Entering edit mode

if it is helpful, I will add the dataset more explicitly:

df <- data.frame(Sample_ID = factor(paste("Sample_", c(1:24), sep ="")),
                 subject = factor(paste("subject_", c(rep(c(1:6), each=4)), sep ="")),
                 condition = factor(paste("condition_", c(rep(c(1:4), times=6)), sep =""))
)

df
   Sample_ID   subject   condition
1   Sample_1 subject_1 condition_1
2   Sample_2 subject_1 condition_2
3   Sample_3 subject_1 condition_3
4   Sample_4 subject_1 condition_4
5   Sample_5 subject_2 condition_1
6   Sample_6 subject_2 condition_2
7   Sample_7 subject_2 condition_3
8   Sample_8 subject_2 condition_4
9   Sample_9 subject_3 condition_1
10 Sample_10 subject_3 condition_2
11 Sample_11 subject_3 condition_3
12 Sample_12 subject_3 condition_4
13 Sample_13 subject_4 condition_1
14 Sample_14 subject_4 condition_2
15 Sample_15 subject_4 condition_3
16 Sample_16 subject_4 condition_4
17 Sample_17 subject_5 condition_1
18 Sample_18 subject_5 condition_2
19 Sample_19 subject_5 condition_3
20 Sample_20 subject_5 condition_4
21 Sample_21 subject_6 condition_1
22 Sample_22 subject_6 condition_2
23 Sample_23 subject_6 condition_3
24 Sample_24 subject_6 condition_4

Is there a solution/feedback on this issue? Thank you in advance!

ADD REPLY • link 20 months ago Luca • 0