How many protein sequences do we usually have for each species?

Question

My question is exactly the one in the title. I add a particular example for more clarity. If I want to study the sequence of protein IkBα, do I find only one sequence or is there a database that contains many sequences from many human host species ?

EDIT: The analysis that I am going to do is to use NLP models to apply it to protein sequences to extrapolate, for instance, protein function knowledge.

Here is a reference of a research paper Learning the protein language: Evolution, structure, and function that considers this application by taking into account the problem of low number of sequences data.

So I think I should have at least 10 protein sequences.

Thank you in advance.

Answer 1

Just to add to Maximilian Press's answer. I happen to know about human immunology and I think Maximilian Press has a good idea.

Genetic studies and immunological studies per se in cellular immunology tend to be restricted within humans. There are limited exceptions to this namely chimpanzees and humanised mammals, most notably humanised mice. In cellular immunology these mammals retain immunologically distinct behaviour but are sufficiently similar to be useful models.

The protein in question is a B-cell repressor, thats more complicated ...

However, using trans-species homology of the protein of interest for

• chimpanzees
• humanised mice
• regular mice

.. represents a good diversity of proteins to use what appears to be a deep-learning function prediction if this was a traditional cellular immunological study. Any immunologist looking at the biological rationale would recognise the model systems deployed.

Method: Blast Generating diversity for given protein is easy, except for humans. Just use online Blast and restrict the search to the species of interest and 20000 hits (?I forget the upper limit) is the maximum and select for more diverse genes. This would work for chimps and (possibly) mice. Thus blast a human gene and restrict the output to the species of interest. Human gene diversity - this much bigger than an online blast could be targeted in accordance with human genetic diversity. The cellular immune system for humans does not conform to background diversity in general. Some geographic human populations do have a HLA predominance but there are no absolute association.

The final option is local blasting the human database for that gene, that is doable and objectively used to derive representative diversity - for this purpose is complicated and would advise against it.

Answer 2

I'm going to start by assuming that you don't actually want 10 protein sequences from the same species, even though you write this in your comment on M__'s (deleted) answer. This will not be a very informative sample of variation in a given protein, except for extreme examples like antibodies.

To get a good sample of variation you almost certainly want to look across species. The paper that you link to does this by using the UniRef database, they then cluster proteins at the 90% similarity threshold. That is well beyond the variation that you'd expect within a species. See their section "Training datasets":

We train our masked language models on a large corpus of protein sequences, UniRef90 (Suzek et al., 2007), retrieved in July 2018. This dataset contains 76,215,872 protein sequences filtered to 90% sequence identity.

You could use an approach like that in the paper to get similar-ish clusters of proteins whose variation you can then analyze. You can use this, but I am somewhat suspicious of that, without reading the paper in too much depth, because we know rather a lot about orthology/paralogy and how it affects function in these cases. Especially in these very elaborate not very introspectable models I worry about what they will do with inadequately curated data.

So instead I'd probably want to first identify the orthology groups (gene families) of interest. These are generally manually curated groupings of proteins.

Once you have the orthology groups that you care about, you can use a database such as KEGG or OMA to identify the specific orthologs (genes) that belong to the group. Then you will have your dataset.