Cuffdiff output:different Ensembl ID but same gene name?

[super0925]

Hi All

I am doing DE analysis on the cow (healthy vs unhealthy).
I align the reads to Cow Ensembl database.
I found that two genes ENSBTAG00000000667 and ENSBTAG00000040244 are both DE but with different LogFC and Q-value.
However, from Cuffdiff result due to the gene.gtf file, the gene names are A6QPR4_BOVIN and A6QQR3_BOVIN repsectively from Cuffdiff result.

But when I want to change them to the Gene Symbol, I found that they are same gene name :'APOL3' .

Pls see the links below:
ENSBTAG00000000667
http://www.ensembl.org/Bos_taurus/Ge...TAT00000000878
ENSBTAG00000040244
http://www.ensembl.org/Bos_taurus/Ge...TAT00000044682

(1)Is this normal? Are they allele?
(2)What shall I do? Shall I leave them both or select the higher logFC ones?
(3)if ENSBTAG00000000667 is DE gene but ENSBTAG00000040244 is not DE, what shall I do?
(4)How about if I align the reads to UCSC genome? Is it better than Ensembl ?

Thank you!

[GenoMax]

They are two copies of apolipoprotein L3 (that are about 55% identical) at two separate locations on the same chromosome. I am not sure how firm the annotation for cow is but they do have separate RefSeq ID's so they must be real.

Your DE result may be a quirk of how the aligner aligned the reads to the common parts of these two genes (did you leave multi-mapping on?)? Did you inspect the alignments to see why one is being called DE and other not?

[super0925]

They are two copies of apolipoprotein L3 (that are about 55% identical) at two separate locations on the same chromosome. I am not sure how firm the annotation for cow is but they do have separate RefSeq ID's so they must be real.

Your DE result may be a quirk of how the aligner aligned the reads to the common parts of these two genes (did you leave multi-mapping on?)? Did you inspect the alignments to see why one is being called DE and other not?

Thank you. Now the two genes are both DE.
In the Ensembl gene annotation file (gene.gtf):
gene_id "ENSBTAG00000000667"; gene_name "A6QPR4_BOVIN",p_id "P3602"
gene_id "ENSBTAG00000040244"; gene_name "A6QQR3_BOVIN", p_id "P26481"

(1)Now I use the default setting of Tophat, it allows the multi-mapping but very little. I don't know how to force multi-mapping off?
And
(2)So now what shall I do? re-alignment or leave the result?
(3)How about use UCSC instead Ensembl?

Thank you!

[GenoMax]

(1)Now I use the default setting of Tophat, it allows the multi-mapping but very little. I don't know how to force multi-mapping off?
And
(2)So now what shall I do? re-alignment or leave the result?
(3)How about use UCSC instead Ensembl?

Thank you!

You could set -g to 1 which will allow only one hit but that could be a random choice between which of the two L3 the read would go to.

UCSC is not going to be any better since the two gene models are also present there.