Tweet
Hello all.
I am currently analyzing some RNA-seq data that we have in our lab, using the tuxedo pipeline.
We are analyzing two biological conditions, where one can be considered normal and the other the "case". We have RNA-seq data from 2 normal samples and 3 case samples. Two of the samples were done in an initial sequencing run, and the remaining three where sequenced in a later run.
I have used Tophat v.2.0.3, and Bowtie 0.12.8 to map the sequences to hg19.
When inspecting the mappings using IGV all seem ok.
Further i ran cuffdiff (v.2.0.0) using the accepted_hits.bam as input, testing all samples against each other using the following command:
cuffdiff --no-update-check --multi-read-correct -q -p 12 --frag-bias-correct $GENOME -M $MASK $REFERENCE /path/n1/accepted_hits.bam /path/n2/accepted_hits.bam /path/c1/accepted_hits.bam /path/c2/accepted_hits.bam /path/c3/accepted_hits.bam
Here the Genome is the path for the genome.fa from hg19.
the mask is a .GTF file of rRNA and mtRNAs, and the Reference is a .GTF hg19 file.
Looking at the output with sample names q1, q2, q3, q4, q5. Only q1, q3 and q4 seems to have successfully been used in cuffdiff differential tests. With values for q2 and q5 always being 0. Also when using cummeRbund and making a boxplot of all genes, q2 and q5 does not have any, while the other three samples show nice boxplots..
I have tried to look for errors in the mapping, but the mappings look nice when inspecting in IGV. Also, when running Cufflinks for each individual sample, they get nice FPKM values.
the two samples that gets no results from Cuffdiff are 2/3 of the second sequencing run...
I have also tried to run cuffdiff comparing only the two samples in question, and they still get 0 for each value (even though cuffdiff reports no errors).
Should i try and map it again using Tophat, or does anyone have any idea what is wrong??
The same commands have been used for all 5 samples, so I don´t see what is wrong with these two..
I am currently analyzing some RNA-seq data that we have in our lab, using the tuxedo pipeline.
We are analyzing two biological conditions, where one can be considered normal and the other the "case". We have RNA-seq data from 2 normal samples and 3 case samples. Two of the samples were done in an initial sequencing run, and the remaining three where sequenced in a later run.
I have used Tophat v.2.0.3, and Bowtie 0.12.8 to map the sequences to hg19.
When inspecting the mappings using IGV all seem ok.
Further i ran cuffdiff (v.2.0.0) using the accepted_hits.bam as input, testing all samples against each other using the following command:
cuffdiff --no-update-check --multi-read-correct -q -p 12 --frag-bias-correct $GENOME -M $MASK $REFERENCE /path/n1/accepted_hits.bam /path/n2/accepted_hits.bam /path/c1/accepted_hits.bam /path/c2/accepted_hits.bam /path/c3/accepted_hits.bam
Here the Genome is the path for the genome.fa from hg19.
the mask is a .GTF file of rRNA and mtRNAs, and the Reference is a .GTF hg19 file.
Looking at the output with sample names q1, q2, q3, q4, q5. Only q1, q3 and q4 seems to have successfully been used in cuffdiff differential tests. With values for q2 and q5 always being 0. Also when using cummeRbund and making a boxplot of all genes, q2 and q5 does not have any, while the other three samples show nice boxplots..
I have tried to look for errors in the mapping, but the mappings look nice when inspecting in IGV. Also, when running Cufflinks for each individual sample, they get nice FPKM values.
the two samples that gets no results from Cuffdiff are 2/3 of the second sequencing run...
I have also tried to run cuffdiff comparing only the two samples in question, and they still get 0 for each value (even though cuffdiff reports no errors).
Should i try and map it again using Tophat, or does anyone have any idea what is wrong??
The same commands have been used for all 5 samples, so I don´t see what is wrong with these two..
, Beginner for RNA-Seq!
Comment