search

wyguo / ThreeDRNAseq

A pipeline for differential expression and differential alternative splicing analysis
https://github.com/wyguo/ThreeDRNAseq
GNU General Public License v3.0
60 stars 27 forks source link

[Help] #14

Closed sk1350 closed 4 years ago

sk1350 commented 4 years ago

Error: need finite 'ylim' values I get this error when trying to plot the mean-variance trend. What does it mean?

Also when trying to run limma it gives me this warning and error: Warning in voom(dge, design, plot = F, span = span, ...) : The experimental design has no replication. Setting weights to 1. Fit a basic linear model ... Fit the contrast model ... [1] "Contrast groups: T13.brep1-T16.brep1; (T10.brep1+T10.brep2)/2-(T13.brep1+T13.brep2)/2; (T13.brep1+T13.brep2)/2-(T16.brep1+T16.brep2)/2" Fit a eBayes model ... Warning: Error in .ebayes: No residual degrees of freedom in linear model fits

Does it mean I cannot use this package with only 2 reps per time point?

wyguo commented 4 years ago

Hi, The 3D App can work for 2 replciates dataset. But from the error message, it seems that your dataset has no replicates. Please double check your input files according to the user manual. I guess it could be your input file problem.

sk1350 commented 4 years ago

Hello, I have used a different dataset that has previously been analysed with DEXSeq. ThreeDRNASeq ran fine up until it gave out the results that stated that there were only 12 DAS genes between two conditions, which is wrong because DEXSeq gave out hundreds of DAS genes. Would you have any idea why that could be?

wyguo commented 4 years ago

Hi, ThreeDRNAseq and DEXSeq are different tools and based on different methods. It is very common that they lead to different results. There are several reasons that you get small DAS gene list by using ThreeDRNAseq

  1. 3D RNA-seq method has a stringent filter for the results of significance. The cut-offs are not only applied to the adjusted p-values, but also applied to the delta PS values.

  2. The reference transcriptome quality has a big inference to the transcript quantification, thereby impacting the quality of alternative splicing analysis.

  3. The RNA-seq data quality has influence to the analysis results.

From: sk1350 [mailto:notifications@github.com] Sent: 22 November 2019 10:09 To: wyguo/ThreeDRNAseq ThreeDRNAseq@noreply.github.com Cc: Wenbin Guo Wenbin.Guo@hutton.ac.uk; Comment comment@noreply.github.com Subject: Re: [wyguo/ThreeDRNAseq] [Help] (#14)

[ External email ] Do not click links or open attachments unless you recognise the sender and know the content is safe

Hello, I have used a different dataset that has previously been analysed with DEXSeq. ThreeDRNASeq ran fine up until it gave out the results that stated that there were only 12 DAS genes between two conditions, which is wrong because DEXSeq gave out hundreds of DAS genes. Would you have any idea why that could be?

— You are receiving this because you commented. Reply to this email directly, view it on GitHubhttps://github.com/wyguo/ThreeDRNAseq/issues/14?email_source=notifications&email_token=AEWMIBUYHZ72CWUURTHASPLQU6VTTA5CNFSM4JJEZFZ2YY3PNVWWK3TUL52HS4DFVREXG43VMVBW63LNMVXHJKTDN5WW2ZLOORPWSZGOEE5FW4A#issuecomment-557472624, or unsubscribehttps://github.com/notifications/unsubscribe-auth/AEWMIBQ5VMGUVDMWRAKSB2DQU6VTTANCNFSM4JJEZFZQ.

The James Hutton Institute is a Scottish charitable company limited by guarantee. Registered in Scotland No. SC374831 Registered Office: The James Hutton Institute, Invergowrie Dundee DD2 5DA. Charity No. SC041796

sk1350 commented 4 years ago

Thank you very much for this explanation.