Hello all,
I'm relatively new to rnaseq analysis. I've been reading a lot and currently trying to come up with a reliable pipeline.
My question is, do you think it's ok to use GATK preprocessed(see below for details) bam files with Cufflinks?
I'll summarize what I've done so far:
I received raw, and clean reads(reads with adaptor sequences, reads in which the percentage of unknown bases (N) is greater than 10%, and reads in which the percentage of bases with quality value ≤ 5 are greater than 50% removed) freon the sequencing facility in fastq format.
Because I didn't want to waste time with preprocessing the raw fq files, I used the clean fqs for alignments(after running FASTQC)
I aligned the reads with Tophat2 using the hg19 references from Illumina iGenomes. I also had to use the --phred64-quals option because the quality scores showed up to be encoded by Illumina v1.5 on FastQC.
My main objective is to perform differential expression and regulation analysis. Before that though, I'm now trying to call variants using GATK.(following the best practices article on Variant Calling on RNAseq posted on the GATK website.
For preprocessing(using the accepted_hits.bam files produced by tophat)
1-I added read group informations, sorted by coordinate, and marked duplicates
2-used the new GATK tool named SplitNCigarReads to split reads into exon segments and hard-clip any sequences overhanging into the intronic regions
3-Performed indel realignment and lastly base recalibration.
Now I'm proceeding with variant calling but later this I'll also get to differential expression analysis.
Back to my question; I do realize that the two processes are for vastly different purposes but do you think I could use final preprocessed bams instead of the original bams produced by tophat for cufflinks? Has anyone tried it/or anything similar?
Other than that I'd also appreciate general suggestions regarding differential expression analysis. My sample sizes are vey low(a group of 4 vs a group of 3 samples) and because they're tumor samples I'm very concerned about the heterogeneity issue. Do you think i should stick with cufflinks or would you suggest some other tool?
Sorry for the long post,
Thanks in advance,
-E
I'm relatively new to rnaseq analysis. I've been reading a lot and currently trying to come up with a reliable pipeline.
My question is, do you think it's ok to use GATK preprocessed(see below for details) bam files with Cufflinks?
I'll summarize what I've done so far:
I received raw, and clean reads(reads with adaptor sequences, reads in which the percentage of unknown bases (N) is greater than 10%, and reads in which the percentage of bases with quality value ≤ 5 are greater than 50% removed) freon the sequencing facility in fastq format.
Because I didn't want to waste time with preprocessing the raw fq files, I used the clean fqs for alignments(after running FASTQC)
I aligned the reads with Tophat2 using the hg19 references from Illumina iGenomes. I also had to use the --phred64-quals option because the quality scores showed up to be encoded by Illumina v1.5 on FastQC.
My main objective is to perform differential expression and regulation analysis. Before that though, I'm now trying to call variants using GATK.(following the best practices article on Variant Calling on RNAseq posted on the GATK website.
For preprocessing(using the accepted_hits.bam files produced by tophat)
1-I added read group informations, sorted by coordinate, and marked duplicates
2-used the new GATK tool named SplitNCigarReads to split reads into exon segments and hard-clip any sequences overhanging into the intronic regions
3-Performed indel realignment and lastly base recalibration.
Now I'm proceeding with variant calling but later this I'll also get to differential expression analysis.
Back to my question; I do realize that the two processes are for vastly different purposes but do you think I could use final preprocessed bams instead of the original bams produced by tophat for cufflinks? Has anyone tried it/or anything similar?
Other than that I'd also appreciate general suggestions regarding differential expression analysis. My sample sizes are vey low(a group of 4 vs a group of 3 samples) and because they're tumor samples I'm very concerned about the heterogeneity issue. Do you think i should stick with cufflinks or would you suggest some other tool?
Sorry for the long post,
Thanks in advance,
-E
Comment