I wonder when to say two proteins are homolog. In the current example I checked the sequence of these two proteins in pfam and as you see they have some shared domains but all domains are not shared. Can I say that these proteins are homolog? How can I be confident about my decision?
$\begingroup$ Did you do a sequence alignment of the two proteins? Homology is often judged on DNA/protein sequence similarity. Other relevant information might be the source species of the two proteins (and their evolutionary distance). $\endgroup$– NicolaiAug 29, 2018 at 14:46
$\begingroup$ It would be better if you can summarize the PFAM results in the form of text. $\endgroup$– WYSIWYGAug 30, 2018 at 7:31
$\begingroup$ I thought that PFAM visuall result better explains my purpose. $\endgroup$– MySkyAug 30, 2018 at 7:58
1$\begingroup$ @MySky It is difficult to search the content shown in the images. You can keep the image but a text summary would be good $\endgroup$– WYSIWYGAug 30, 2018 at 8:05
$\begingroup$ Worth mentioning that this was cross-posted to BioStars. $\endgroup$– Joe HealeyAug 30, 2018 at 9:41
Homology means shared evolutionary ancestry. Sequence similarity is often used as a proxy for homology but inferences should be made with care.
The similarity between two genes/proteins should not just be good but has to be statistically significant (metrics like E-value) for the two genes/proteins to be considered homologous.
INFERRING HOMOLOGY FROM SIMILARITY
The concept of homology – common evolutionary ancestry – is central to computational analyses of protein and DNA sequences, but the link between similarity and homology is often misunderstood. We infer homology when two sequences or structures share more similarity than would be expected by chance; when excess similarity is observed, the simplest explanation for that excess is that the two sequences did not arise independently, they arose from a common ancestor. Common ancestry explains excess similarity (other explanations require similar structures to arise independently); thus excess similarity implies common ancestry.
However, homologous sequences do not always share significant sequence similarity; there are thousands of homologous protein alignments that are not significant, but are clearly homologous based on statistically significant structural similarity or strong sequence similarity to an intermediate sequence. Thus, when a similarity search finds a statistically significant match, we can confidently infer that the two sequences are homologous; but if no statistically significant match is found in a database, we cannot be certain that no homologs are present.
Members of a protein family are descendants of a common ancestor and are hence homologous. However, in the course of evolution they would have acquired new domains or reshuffled their domains such that their sequences are no longer similar. Proteins that have full length sequence similarity are called homeomorphic (Wu et al., 2004). Therefore, members of a protein family may be homologous but not homeomorphic. However, homeomorphic proteins can evolve independently and therefore may not be considered homologous.
Identifying homologous proteins is, therefore, not a simple task. Machine learning algorithms are used for better identification of homologous proteins. Some of these algorithms are mentioned in the linked papers.
In general, global similarity, rather than local similarity should be considered for identifying homeomorphs. See What is the difference between local and global sequence alignments?
1$\begingroup$ This is the definition of homology that disturbed me. you say we must consider global similarity rather than local similarity. accordingly, we can't say these two proteins are homologs, because just about 60% of their lengths have significant (above 80%) similarity and the rest of their sequence is not similar. But actually, we see that they share 4 exactly similar domains, and have 3 nonsimilar domains. according to shared domains we can say those are homolog (at least in part of their sequence). I am really confused. $\endgroup$ Aug 30, 2018 at 7:55
$\begingroup$ Wikipedia (Protein family): Proteins in a family descend from a common ancestor (see homology) and typically have similar three-dimensional structures, functions, and significant sequence similarity. The most important of these is sequence similarity (usually amino acid sequence) since it is the strictest indicator of homology and therefore the clearest indicator of common ancestry. Proteins that do not share a common ancestor are very unlikely to show statistically significant sequence similarity. $\endgroup$ Aug 30, 2018 at 8:39
$\begingroup$ Could you please lead me to some good papers that give me a deeper understanding of the definition and concept of homology? As I previously said, I am really confused about this. $\endgroup$ Aug 30, 2018 at 8:42
$\begingroup$ I have updated the answer @MySky $\endgroup$– WYSIWYGAug 30, 2018 at 9:02
Domain is a module. Proteins are composed of one or more domains. Proteins are modular. Homology describes the relationship between domains, because domains are the primary evolutionary units. It is possible for two proteins to be homologous in one and non-homologous in another domain.
For example, alpha/beta-hydrolases (Ollis et al., 1992, Protein Eng. 5: 197-211) belonging to different families are homologous in the alpha/beta-hydrolase domain, but non-homologous in the "cap" domain.