Crucial difference between dexseq and HTSeq-count

Question

[igolkinaanna11]

Hi everyone!

I decided to compare exon counts (DEXseq) and gene counts (HTSeq-count) for my alignments (done by STAR). Annotation(gtf) for mapping is just the same as that for counting. .gtf for DEXseq (modified by dexseq_prepare_annotation.py) a correspond to .gtf for HTSeq.

So everything seems ok, but I cannot interpret inconsistencies results. Under my hypothesis total count of exons should be somehow correlated with gene count. But what I see:

dexseq:

ENSG00000000003:001    248
ENSG00000000003:002    86
ENSG00000000003:003    63
ENSG00000000003:004    56
ENSG00000000003:005    66
ENSG00000000003:006    105
ENSG00000000003:007    91
ENSG00000000003:008    82
ENSG00000000003:009    80
ENSG00000000003:010    0
ENSG00000000003:011    16
ENSG00000000003:012    5
ENSG00000000003:013    0
ENSG00000000003:014    1
ENSG00000000003:015    2
ENSG00000000005:001    0
ENSG00000000005:002    0
ENSG00000000005:003    0
ENSG00000000005:004    0
ENSG00000000005:005    0
ENSG00000000005:006    0
ENSG00000000005:007    0
ENSG00000000005:008    0
ENSG00000000005:009    2

HTSeq-count:

ENSG00000000003    2
ENSG00000000005    2

Why it is too small for ENSG00000000003 in HTSeq-count?

Thank you for answers,

Anna

[Martombo]

to have a better picture I would suggest you to plot all the gene vs exon counts in your data. that will tell you if what you observe for the first two genes is true for a vast amount of them or not. also, what is the total of the counts that you get in the two analyses? is it comparable?

[igolkinaanna11]

I've done such plot (gene count vs total exon count per gene) and there is no correlation so I have the same situation is for vast majority of genes.

Additional info from dexseq:

_ambiguous    752401
_empty    16077636
_lowaqual    0
_notaligned    0

Additional info from HTSeq-count:

__no_feature    40579506
__ambiguous    44464
__too_low_aQual    0
__not_aligned    0
__alignment_not_unique    23644541

Total counts form dexseq: 56934303

Total counts from HTSeq: 65396229

[rohitsatyam102]

Hi !

Were you able to figure this out? I am also trying to understand in what mode does dexseq python scripts runs HTseq in the background?