Pseudogene ontology

I’m currently working on reasoning with ontologies, and found in big
interest in reading your paper about reasoning of the pseudogene set.
I’m particularly interested in the time you give for the reasoning to
complete, since performance is my biggest issue. So I wanted to ask
you what was the size of the data you used, since the pseudogene set
does not seem to be available online anymore.

The pseudogene data set used in this paper can be found at:, which is based on
Ensembl genome release 48. There are 2,294 duplicated pseudogenes and
10,187 processed pseudogenes, in chromosome 1-22, X and Y. Hope this

PGOHUM00000250823 and SETP13/SETP3 pseudogenes


Could you please investigate the support you have for PGOHUM00000250823? At NCBI, we have PGOHUM00000250823 associated with HGNCid 42932, official symbol SETP13, and RefSeq accession NG_032538.1. However, we have a nearly identical RefSeq accession NG_032022.1 associated with HGNCid 31115 and official symbol SETP3. On the current human reference assembly, GRCh38, both NG_032538.1 and NG_032022.1 align perfectly to the same locus and have no other hits of comparable quality to the reference assembly (or to alternate assemblies HuRef or CHM1_1.1). In NCBI’s latest annotation (Annotation Release 106) SETP3 was annotated on the assembly but SETP13 was not because it overlapped with SETP3. Do you have any evidence these are distinct pseudogenes? If not, NCBI’s preference would be to preserve the older nomenclature associated with SETP3 and NG_032022.1. Also, if we agree that SETP13 is redundant with SETP3 then I will proceed to notify HGNC and will CC you on that email.

I looked at the PGOHUM00000250823/SETP13 locus. Our pipeline
predicted it as a pseudogene to SET with around 90% sequence identity.
When compared to SETP3 locus, SETP13 lacks sequences at both 3’ and 5’
ends, which can be aligned to the UTR regions of SET. Our pipeline
missed these sequences at both ends because it searches for homologous
sequence to CDS regions only. We are actually thinking of including
some checks of UTR alignment in our revised pipeline. Thanks for
pointing this case to us.

We have no problem to merge the SETP13 locus with SETP3.

Questions regarding


I am trying to get fly pseudogene information available from
I want to know the parent gene of any pseudogene. provides “parent proteins”, such as FBpp0112526. However, I cannot find this id in flyabase. Is it the Flybase ID? If not, what database ID is that from?

The fly pseudogene information currently available on website is old. As you can see it is from Ensembl build 50, when the current Ensembl release is 75. The FBpp00… id is an Ensembl protein ID based on flybase. However a lot of these ID have been deprecated between the two releases. We are currently preparing a new annotation file for fly pseudogenes based on the final stable gene annotation and it is going to be available online shortly. However if you still want to use the fly pseudogene annotation you can parse all the parents protein ids in the file using Ensembl biomart and you can see which ids are still current and which are retired. Also the Ensembl biomart gives you the option to get the corresponding transcript and gene id for each protein id.

By downloading the fly pseudogenes from, I can get >1000 pseudogenes, but if I use BioMart, after selecting pseudogene, I can only get 175 pseudogenes. Why?

Since all the pseudogenes at were identified by your lab, you must have their parent information (gene name or transcript name). Could you provide that information? I do not need parent protein name.

By the way, what pipeline did the lab use to identify the pseudogene? The pseudogene has UTR? Which paper did the lab publish regarding how the pseudogene was identified?

As I said before the fly pseudogenes that are available from are based on a very old gene annotation (Ensembl 50). The quality of the pseudogene annotation is dependent on the quality of the gene annotation. As such, since the fly gene annotation for Ensembl build 50 was just a draft, many of the pseudogene entries that we obtained from build 50 are actually false positives. Currently we are working on the latest fly pseudogene annotation and we’ll make it available soon (next couple of weeks). In our latest annotation we have about 150 pseudogenes. This last set was obtained using a combined manual and automatic annotation. The automatic annotation was obtained using PseudoPipe – a pseudogene annotation pipeline.

Also if you select pseudogenes in BioMart, you will find only the Ensembl annotated pseudogenes. Those pseudogenes were identified using the Ensembl annotation pipeline.

Gerstein lab has published numerous papers regarding pseudogene annotation. For the full list please see:

The pseudogenes do have UTR, however at the moment we do not provide an UTR annotation for fly pseudogenes.

How many of your 150 psudogenes are in the 175 psudogenes in Ensemble obtained via BioMart?

Your pseudogene pipeline starts with protein sequence, and that’s why your report has no UTR?

I attach here (see below) the latest fly pseudogene annotation.
Regarding your questions:

1. There is a reasonable overlap between Ensembl pseudogenes and our set. However I have to mention that Ensembl pseudogene are based on the automatic annotations while our pseudogenes are also manually annotated.

2. yes, our pseudogene annotation pipeline uses the protein data information.

Thanks for that. But the attached file only contains the common ones between Gerstein lab annotation and Ensembl annotation? since each row has a Ensembl ID.

The file contains the latest Gersteinlab annotation. Our annotation was done using a combination of automatic and manual annotation so it is of higher quality than the Ensembl one. The pseudogenes do have Ensembl IDs for easier processing.

I am confused. Could you, for example, show me one pseudogene that is annotated by Gerstein lab, but not by Ensembl?

Maybe I was not clear, our pseudogenes are available through Ensembl, but there are Ensembl-only pseudogenes that have no correspondent in our data set. Also we define their biotype while in Ensembl you won’t find the biotype information.

oh? I heard Gerstein lab just submitted the latest annotation not long ago , and the latest annotation will not be available to public right now. your previous attached file is exactly the latest one that hasn’t been published?

The file you sent me was obtained by BioMart of Ensembl? if so , how to set the "Filters" there in order to get the same file as you.

Could I also have the pseudogenes that are not included in Ensembl pseudogene list?

By the way, what is "processed_pseudogene" vs "unprocessed_pseudogene" ?

Sorry for keeping bothering you and thanks for your patience.

The file that I sent you is our latest and yet unpunished annotation and yes it is not publicly available at the moment. But this will be the official list of pseudogene to use for the fly genome since it is a high quality set, each pseudogene annotation being validated through manual inspection.

The “processed” and “unprocessed” nomenclature refers to the pseudogene biotype, a classification of pseudogenes based on their mode of creation (e.g. processed pseudogenes were formed through retrotransposition while unprocessed pseudogenes are usually the product of duplication). If there is no defined nomenclature , e.g. just “pseudogene” in the biotype field, that means we could not assign a definite biotype to that particular element.

If you want to compare our pseudogene set with the one from Ensembl I would recommend you to use bed tool. Create a bed file for each set and intersect them.

Pseudogenes lacking on short arm of chr13, 14, 15 and 22

Found your paper ( about
pseudogene’s very informative.

Have been looking at the distribution of pseudogenes(
using the Ensemble annotation as well as that found on

However, pseudogenes seem to be lacking on short arm of chr13, 14, 15 and 22.

Could you please let me know if this is a known biological pattern or
some missing annotation?

In human, chromosomes 13, 14, 15, 21 and 22 are acrocentric. They are made of a very long arm and a very short arm that is homologous across the five chromosomes. Only the long arm has been sequenced and annotated. This is way there are no pseudogenes (or genes for that matter of fact) annotated on those chromosome arms.

For more info please see:

Question about data

I have a question about database.
I’m analyzing human pseudogene database and noticed that many "processed" pseudogene (>70%) don’t have polyA.
It seems like opposite of what textbook says. Is that true?
What’s the criteria of "processed pseudogene" in

I came to find another question.
I tried to blat search using several pseudogene sequence from each class of "polyA: "0" or "1" or "2" or "3" ".
But most of PolyA class 1,2,3 don’t have convincing polyA tail compare to following criteria.

Polya: "0" or "1" or "2" or "3".
"0":no polyA tail (> 30 A in 50 bp window) detected of the pseudogene
"1" : has polyA tail and also polyadenylation signal with 50 bp of the begining of the tail
"2" : has polyA tail and polyadenylation signal within 50-100 bp of the begining of the tail
"3": has polyA tail but no polyadenylation detected.

Does number coordinate of data depends on human genome assembly GRCh37/hg19?

Pseudogenes are identified primarily by homology matching of protein sequence against the human genome. However, the pipeline that we use incorporates poly A analysis. Our group published a paper a few years ago where we showed that ~ 50% of ribosomal protein pseudogenes do not have a detectable poly A signal. . We believe that this is due to decay in genome sequence and nucleotide substitutions.

For detecting poly A signals and classification, the following criteria is used according to the paper linked above.

We searched a 1000-bp region that was 3′ to the pseudogene homology segment, with a sliding window of 50 nucleotides for a region of elevated polyadenine content (>30 bp), and picked the most adenine-rich 50-bp segment as the most likely candidate. An interval of 1000 nucleotides was used because of the possible existence of 3′-untranslated regions (3′-UTRs); 90% of 3′-UTRs are of length less than 942 bp (Makalowski et al. 1996). In addition, we searched in the same 1000-bp region for candidate AATAAA or other polyadenylation signals and checked whether they were upstream of the candidate polyadenine tail site.

This criteria might not be very stringent.

And yes, the pseudogene coordinates are dependent on the human genome from which it is derived, hence the human genome version number is important.

Search help for PseudoPipe program


Recently, I have read your
published paper named" PseudoPipe: an automated pseudogene identification
pipeline"( Vol. 22 no. 12 2006, pages
1437–1439/doi:10.1093/bioinformatics/btl116), which impressed me so much. I
really admire your and co-workers’ excellent work.
After reading the literature, I downloaded the PseudoPipe program(Pipeline
Source Code) at and tried to use it to identify
pseudogene sequences in mammalian genome.But there are some questions
during pre-experiment.I input the exsiting data
(caenorhabditis_elegans_62_220a) and installed Python 2.26,howeverthe
PseudoPipe program failed to run the analysed the
fastaAlign did not do well.It really puzzle me a lot, and I will appreciate
it if you can solve them for me.

Please note that PseudoPipe was written to discover pseudogenes in mammalian genomes, it does not work well in C. elegans.

The pseudogene information of zebrafish in should be updated

In, the pseudogene datasets of zebrafish (Danio rerio)
was based on old annotations (Ensembl 55?). There were about ~1800
processed pseudogenes. However, based on a recent research
there were rare pseudogenes in zebrafish. (Only 21 processed
pseudogenes, according to Supplementary Table 14 in the published

Is this great conflict due to the old annotations?

You are right about the zebrafish pseudogenes in the The results were based on an old genome assembly, ENSEMBL release 55, which was done in year 2009. We do notice that the pseudogene number is way too high, which we believe partially due to the quality of the genome assembly, and partially due to the reason that the pipeline parameters were optimized with primates. Given that, for up-to-date pseudogene information in Zebrafish, people should refer to the Nature publication:

Thanks for pointing this out to us.

information about


I am interested in using the information in your database to design PCR probes that would recognize usable and ensuing pseudogenes for several genes.

Do I need to obtain any type of written permission to use this information?


I checked one gene, with 9 pseudogenes listed and tried to align the
sequences to make PCR primers to detect the 10 copies, however, I realized
that being a bit naïve about pseudogenes led me down the wrong path, as I
thought the sequences would be more similar, and adept to being used to
estimate copy number for inserting foreign genes. While I did get regions
that hit 3-6 of the 10 genes, it wasn’t consistent enough.

I was wondering if you have the data about % conservation or any types of
algorithms that would predict the % conservation of pseudogene to gene and
pull out those names/gene Ids and number of pseudogenes?

It would be helpful if you can tell us a bit more about what you are trying to do.
I assume you are looking at human pseudogenes. We do have percent identity between the parent protein and the pseudogene.

I’m trying to figure out a sensible way to use the numbers of the pseudogene/gene as a natural standard curve for real time PCR. See attached excel file. I chose at random genes with 9 to1 listed pseudogene which theoretically would allow me to target endogenous genes of different copy number and get some type of standard curve. This is assuming equal efficiency etc.

I didn’t pay attention to the column "Identity" but now I’m thinking I can sort out genes based on high identity and try again?

I think that identity should be taken into account when you are creating the standard curse. Also, note that in the excel file, there is a column of fraction (after gene ID), which indicates the fraction of a parent gene aligned to its pseudogene. The start and end coordinates of an alignment are also in the excel file (columns between protein ID and gene ID). Maybe you want to take these into consideration too.

Pseudogene minilist for PCR.xlsx

Comparing chromatin state analysis at pseudogene regions

I am very interested to compare our chromatin state analysis at the pseudogene regions. I found this file at your website:

Could you please let me know if this is the right place to compare? I saw you do have h1-esc there. If I understand correctly, you classified each pseudogene as being in either active (1) or silent state (0).

The chromatin state, promoter prediction and pol2 binding regarding to pseudogenes in H1-hesc are included in the psiDR file. Please let me know if you have any questions about that file.