Entering edit mode
2.8 years ago
salmon
▴
10
Hello all,
I performed bbmap on my RNA-seq paired sequence data using following cmd
bbmap.sh in1=J2_R1.fastq in2=J2_R2.fastq out=output_J2.sam ref=im4.fasta nodisk
The header of generated sam file is
@HD VN:1.4 SO:unsorted
@SQ SN:k141_1006 LN:2503
@SQ SN:k141_5512 LN:5393
@SQ SN:k141_4772 LN:4387
@SQ SN:k141_3267 LN:4531
I then used htseq-count using following cmd
htseq-count -s no -a 10 output_J2.sam im4.gff > J2.count
The J2.count shows following
__no_feature 68032
__ambiguous 0
__too_low_aQual 45285
__not_aligned 21494716
__alignment_not_unique 0
The number of features are very less as compared to reads not aligned. Is it due to less mapping during bbmap (0.52%) Any help highly appreciated!
Thanks
before we go any further, do check if the sequence names are the same between the gff file and the BBmap output file (or thus the fasta reference file)
wait, do you mean that you only have a 0,52% mapping rate from BBmap ? If so, then yes likely that is the cause of those low counts you see (== practically no alignments thus)
Thank you for your reply. Yes mapping from bbmap shows only 0.52%.
well, there you go then ... (== is the issue of the low counts) so nothing htseq-count messed up thus.
so we moved the issue on to why BBmap is only mapping 0,52% ?
are you using a transcriptome as reference?
you are using the correct reference for those samples to map to?
and not really a solution for this but just some advice: you might consider changing htseq-count for FeatureCount (more robust, quicker, ... )
As lieven.sterck said tell us what you are aligning to i.e. type of reference. Do you need to account for splicing? If you do then you need to add relevant options for
maxindel=NNNNNandintronlen=NN. Are reads in R1/R2 files in sync? Did you keep them together during trimming (if you did trim). If you havesamtoolsin your$PATHthen you can directly write BAM files (useout=file.bam), no need to use SAM files.Can you show us a sample of your fasta headers?
grep ">" im4.fasta | head 4. If there are spaces in your fasta headers they will be passed on to the BAM file. To avoid that you may want to add optiontrd=tat time of alignment.I am using reference downloaded from Patric after performing genome annotation. Genome annotation was performed on the DNA sequence of the same bacteria (for which I am trying to do RNA seq analysis)
Reads R1/R2 are for bacterial RNA sequence samples. I made sure they are aligned well while performing adapter trimming and contaminant filtration. Yes I kept them together during trimming.
I tried to use BAM file. But when I am trying to see the header of it is shows encrypted message like following. So i tried using SAM file. I am not sure why I am getting these encrypted codes.
head -5 im4.fasta shows following. It shows no spaces.
head -5 im4.gff shows:
yes, you can't just open a BAM file with cat or a text editor or head or such, it is a binary format (aka not human readable thus). You can still do so but then you'll need to go through that samtools tool which can 'translate' it on the fly to human readable format.
Keeping files in BAM format has a huge storage advantage, they are much smaller than their SAM counterparts. So advice is to keep those files as much as possible in BAM format. (just a side node)
The head of your fasta file looks a bit strange. Can you confirm that the sequence starts on the same line as the header? (or is this a copy-paste typo when posting?) If it is on the same line: that's not valid fasta format (well technically it is but it just will not contain all of the sequence you might think it does)
There was a typo for the fasta format of Im4. I updated it as the reads starts from different line.
Also, one thing to mention when I checked the taxonomy (Through GTDB-TK and PATRIC) of bin (that was an input to PATRIC for genome annotation), it showed as Pseudomonas (gammaproteobacteria). But my pure culture sample belongs to different class (betaproteobacteria) and species. I referred this paper where they showed that Pseudomonas is phylo-genetically related to my bacteria. Still, I think due to unrelated taxonomy, my mapping percentage is very low. But I am not sure how to proceed further. I request if you can please help me.
Thank you so much
that is of course also an option , that your sample is contaminated (or a sample mix-up). That would indeed also be reflected in the very low mapping percentage . If I would map a plant sample against human genome it will also reflect in very low mapping percentages.
Just one question though as I am using the annotated gff file generated by PATRIC. It shows the featureID as fig|270368.6.peg.1, fig|270368.6.peg.2, fig|270368.6.peg.3 ...so on for each contig ID. I tried to convert these each feature ID to gene ID using ID mapper but I am unable to. Do you think low mapping might be due to these feature ID names.