No announcement yet.
  • Filter
  • Time
  • Show
Clear All
new posts

  • HTseq: Very few counts recognised

    Ive seen a lot of threads on this, but I can't figure it out. I got 16-60 millions single end reads in each library. Ive used Tophat 2 with UCSC GTF file for hg19.

    This is my code:

    samtools view accepted_hits.bam | \
    htseq-count -m intersection-nonempty -s no -a 10 \
    - UCSC/hg19/genes.gtf \
    > Out.txt

    Here is a typical result, its propotional to the library size:

    no_feature 7013689
    ambiguous 269370
    too_low_aQual 0
    not_aligned 0
    alignment_not_unique 6645341

    How come i get on average 25 - 50% reads that is "no_feature",
    "ambiguous" or "alignment_not_unique".

    This is RNAseq, and if I must visually inspect, how to precede?

  • #2
    Perhaps you have a lot of immature mRNAs or a lot of expressed repeat regions. The general idea is to look at some of the alignments in IGV and see if they really don't match anything. Also ensure that the chromosome names in the BAM file and GTF file match (that probably causes this sort of thing half the time).


    • #3
      Hi again!

      I have now tested HTSeq with all modes, also upgraded to Python 2.7.6 and inspected using IGV.

      Click image for larger version

Name:	Screen Shot 2014-01-20 at 22.43.27.png
Views:	1
Size:	17.1 KB
ID:	304414

      Here is in total 4 reads, one with mapping quality 50 and three with 3.
      I used HTSeq option -a 0, so they should been picked up..

      All three modes only counts 1 read.. How can this be?


      • #4
        HTSeq-count also looks at the NH auxiliary tag. With a MAPQ of 3, it's likely that three of those are multimappers (this will be the case if you used tophat2) and would be (properly) ignored.


        • #5
          Oh yeah, that make sense.

          How come you know so much about everything? Where have you learned?

          But, its pretty sure something is wrong here right, so I should keep looking? I have checked my GTF, the chromosome names are the same.


          • #6
            I have made a SAM file with -samout option and checked around a bit..

            From Tophat.log I get 21.8 mill. total kept reads
            In my SAM I get a total of: 20.90 mill., wonder where the rest are?

            Of the 20.9 mill. I have:

            17.7 mill. NH:i:1 of which 158.000 ambiguous
            I also get 3.4 mill. alignment_not_unique &
            1.6 mill. no_feature

            Looking at the HTSeq output file I get:

            no_feature: 3.8 mill.
            ambiguous: 158.000
            alignment_not_unique: 3.4 mill.

            So the SAM has 1 million reads less than the BAM. Also "no_feature" is different in the SAM and HTSeq output..

            I tried to watch specific reads in IGV, but selecting reads by read name (right click the BAM track and choose "select by name", does not change the view....Annoying).

            But anyone have something to add on this?


            Latest Articles


            • seqadmin
              Advanced Methods for the Detection of Infectious Disease
              by seqadmin

              The recent pandemic caused worldwide health, economic, and social disruptions with its reverberations still felt today. A key takeaway from this event is the need for accurate and accessible tools for detecting and tracking infectious diseases. Timely identification is essential for early intervention, managing outbreaks, and preventing their spread. This article reviews several valuable tools employed in the detection and surveillance of infectious diseases.
              Yesterday, 01:15 PM
            • seqadmin
              Strategies for Investigating the Microbiome
              by seqadmin

              Microbiome research has led to the discovery of important connections to human and environmental health. Sequencing has become a core investigational tool in microbiome research, a subject that we covered during a recent webinar. Our expert speakers shared a number of advancements including improved experimental workflows, research involving transmission dynamics, and invaluable analysis resources. This article recaps their informative presentations, offering insights...
              11-09-2023, 07:02 AM





            Topics Statistics Last Post
            Started by seqadmin, Yesterday, 08:12 AM
            0 responses
            Last Post seqadmin  
            Started by seqadmin, 11-22-2023, 09:29 AM
            1 response
            Last Post VilliamPast  
            Started by seqadmin, 11-22-2023, 08:53 AM
            0 responses
            Last Post seqadmin  
            Started by seqadmin, 11-21-2023, 08:24 AM
            0 responses
            Last Post seqadmin