6
$\begingroup$

Humans alone have thousands of proteins. With that in mind, it seems like the total number of proteins among all species would be very large.

Are there any available estimates for how many proteins exist on earth in all organisms? I'd also be interested in how many of these are unique proteins as opposed to proteins that are very similar to other proteins, i.e an estimate of non-redundant proteins alongside the redundant proteins.

$\endgroup$

closed as too broad by David, another 'Homo sapien', kmm, mdperry, March Ho May 17 '17 at 18:04

Please edit the question to limit it to a specific problem with enough detail to identify an adequate answer. Avoid asking multiple distinct questions at once. See the How to Ask page for help clarifying this question. If this question can be reworded to fit the rules in the help center, please edit the question.

  • 2
    $\begingroup$ Well, remember that just about all proteins are evolved from other, related proteins, so how many you consider unique is really a matter of definition. If you and I have a protein with a single amino acid substitution, is that a new protein? Kind of makes the question impossible to answer. $\endgroup$ – Bryan Krause Apr 25 '17 at 4:23
  • 1
    $\begingroup$ As @BryanKrause mentioned, if you intend 'unique' to signify "found only in humans/only inone of all other organisms", there will be very few. If you mean 'unique' to mean "different", we don't have that answer. We're still discovering new species almost every day! $\endgroup$ – anongoodnurse Apr 25 '17 at 4:48
  • $\begingroup$ @BryanKrause Protein redundancy is a fairly common method of allowing clusters of proteins to be represented by a single protein. This avoids biases introduced by overstudied systems or by large groups of very similar proteins masking distant proteins that would have different biochemistry. It's a scientific way of generating lists of "unique" proteins. $\endgroup$ – James Apr 25 '17 at 6:36
  • $\begingroup$ I see no reason it should be closed as unclear. It's a simple easy to understand the question. It's subjective indeed, but there are papers estimating these types of questions, so it's not completely a matter of opinion. $\endgroup$ – James Apr 25 '17 at 7:04
  • $\begingroup$ I'm voting to reopen this question since there have been scientific studies on exactly this question. See "Towards the completion of the earth's proteome" Perez-Iratxeta et al. 2007. This question can be quantified and objectively explored. On the other hand, the unclear votes appear to be from a miscommunication that single sequences can represent homologous clusters. I've edited the question to make this part of the question more clear. $\endgroup$ – James Apr 25 '17 at 9:11
6
$\begingroup$

Current records

According to Uniprot, there are 85,381,808 protein records, and with the UniRef90 filter (i.e removing records that can be represented by an entry with at least 90% sequence similarity), there are 42,424,511. However, these databases are moving targets and will change over time. We will sequence more species, find novel splice isoforms, and various other methods will expand the databases. Indeed, the databases will be truncated also from time to time as some hypothetical proteins may be based on genes that turn out to not code for proteins after all.

In 2007, a study estimated that the earth's proteome would contain around 5 million sequences, and that the majority of these would be elucidated by 2012. I suspect this is a very thorough study, however, a lot has changed in the last 10 years. This estimate is actually less than the nearly 9 million species estimated in more recent studies.

Approximate estimate

So let's do some back of the envelope maths. Let's assume the article estimating nearly 9 million species is about right, and that we've only catalogued some 1.2 million. But UniProt isn't even close to this number.UniProt contains 25477 scientific names in it's controlled vocabulary. So for 25 thousand names, we have 85 million protein records. What if we had 8.75 million names? Let's assume:

$\frac{Predicted~Proteins}{Predicted~Species}=\frac{Known~Proteins}{Known~Species}$

We can rearrange this to:

$\frac{Predicted~Species~\times~{Known~Proteins}}{Known~Species}= Predicted~Proteins$

Generous estimate (Uniprot, 335527 proteins per species):

$\frac{8750000\times{85381808}}{25477}=2.932413e^{+10}$

Conservative estimate (Swissprot, 41 proteins per species):

$\frac{8750000\times{554241}}{13408}=3.616952e^{+7}$

For the sake of completeness let's assume the number of <90% identical proteins will remain around half that value. We can say that there might be around $1.8e^{+7}$ to $1.5{e}^{+10}$ "unique" proteins, less than a trillion ($1e^{+12}$). Given the absurdly generous 335 thousand proteins and very stingy 41 proteins per species, we can be fairly certain that if there are indeed 8.75 million species, the number of proteins will fall between those estimates.

The biggest assumption here is that the proteins have a linear relationship with species which is unlikely to be the case, and at the generous estimate we are pretending that there are no proteins in UniProt that don't have species annotation. As for Swissprot, this only includes proteins that have been manually curated, so this ignores many proteins that are safe to assume exist and typically only covers proteins that are of interest to scientists.


A minor correction to your question, the UniProt lists ~20 thousand protein coding genes in the human proteome, not millions. Those protein coding genes are subject to various post translational modifications and isoform splicing, so there will be more final proteins than 20k.

$\endgroup$
  • $\begingroup$ The human proteome can contain millions of different proteins if we consider proteins with different post translational modifications (PTMs) to be different! $\endgroup$ – Jeppe Nielsen Apr 25 '17 at 11:42
  • 2
    $\begingroup$ Just because a database does not consider proteins with different PTMs to be different, does not mean that the proteins are identical. If you disregard PTMs you cannot explain the ABO bloodtypes as they are caused by differences in the glycosylation patterns. $\endgroup$ – Jeppe Nielsen Apr 26 '17 at 1:08
  • 1
    $\begingroup$ Another good example would be the insulin receptor, which comes in many different variants. $\endgroup$ – Jeppe Nielsen Apr 26 '17 at 1:10
  • 1
    $\begingroup$ So you would say that the genome contains ~20,000 protein coding genes. The proteome is much larger. $\endgroup$ – canadianer Apr 26 '17 at 18:03
  • 1
    $\begingroup$ Each UniProt record contains isoforms and modifications. That they choose to organize their database this way doesn't seem to me to have any bearing on the definition of the term proteome. Any definition I have heard refers to the entire complement of expressed proteins. See UniProt's, for example: A proteome is the set of protein sequences that can be derived by translation of all protein coding genes of a completely sequenced genome, including alternative products such as splice variants... $\endgroup$ – canadianer Apr 27 '17 at 1:06

Not the answer you're looking for? Browse other questions tagged or ask your own question.