The sequences are real, not predicted. It is the function of the sequences, the annotation of the gene, that is predicted. Whether a gene is annotated in any database such as GenBank as "predicted" or simply as "gene X encoding protein Y" is arbitrary. Some annotators will freely label any section of a new genome with x percent identity to genes annotated in other species as "gene X" and others will be more conservative and only state that this region of the genome is x percent similar to regions of other species genomes. In modern genomic biology the true function of a region of a genome is almost never experimentally determined,. However, vast experience has shown that most regions of most genomes with high sequence similarity also turn out to have similar functions. There is room for plenty of error in plenty of special situations, such as gene duplication events, sparse data sampling, and horizontal transfer of genes. Which of the dozens of tyrosine kinases in a human is "the homolog" of one specific copy of the dozens of tyrosine kinases in a yeast that is being studied experimentally? This may not be possible to determine for sure. However, it is much easier to infer shared ancestry between copies of tyrosine kinases in human, goriilla, chimpanzee and orangutan based on genomic location and other clues which have been erased by the distance between human and yeast.

So, whether or not to use genes that have been annotated as "inferred" or even the genes that have been fully annotated, is a judgement call that should be based on a lot more knowledge about the gene being investigated. Is it most likely to be single copy or one of a set of many duplicates? Are the species with "predicted" annotation closely related or distant, based both on this gene and others? Does the phylogeny built with this gene tend to agree with that built from ribosomal genes, DNA polymerases, or other genes?

Inferred homology by sequence similarity really has nothing to do with what annotaitons (predicted or otherwise) is assigned to any given sequence. If BLAST gives you strong evidence of homology, then you can use the sequence. Imagine if it had simply been deposited into GenBank with no annotation at all - just raw genomic sequence reads. Would you then dismiss it as valid homology hits because it had no annotation assigned?

The sequences are real, not predicted. It is the function of the sequences, the annotation of the gene, that is predicted. Whether a gene is annotated in any database such as GenBank as "predicted" or simply as "gene X encoding protein Y" is arbitrary. Some annotators will freely label any section of a new genome with x percent identity to genes annotated in other species as "gene X" and others will be more conservative and only state that this region of the genome is x percent similar to regions of other species genomes. In modern genomic biology the true function of a region of a genome is almost never experimentally determined,. However, vast experience has shown that most regions of most genomes with high sequence similarity also turn out to have similar functions. There is room for plenty of error in plenty of special situations, such as gene duplication events, sparse data sampling, and horizontal transfer of genes. Which of the dozens of tyrosine kinases in a human is "the homolog" of one specific copy of the dozens of tyrosine kinases in a yeast that is being studied experimentally? This may not be possible to determine for sure. However, it is much easier to infer shared ancestry between copies of tyrosine kinases in human, goriilla, chimpanzee and orangutan based on genomic location and other clues which have been erased by the distance between human and yeast.

So, whether or not to use genes that have been annotated as "inferred" or even the genes that have been fully annotated, is a judgement call that should be based on a lot more knowledge about the gene being investigated. Is it most likely to be single copy or one of a set of many duplicates? Are the species with "predicted" annotation closely related or distant, based both on this gene and others? Does the phylogeny built with this gene tend to agree with that built from ribosomal genes, DNA polymerases, or other genes?

As pointed out by several people, it is the % identity among sequences that counts for phylogenetic analysis, not whether that particular sequence corresponds to a well characterised gene or not. Including such predicted genes into an analysis could strengthen the case for them corresponding to real genes, and maybe sooner or later even assign a function to them, or confirm the existence of a putative gene product that they would encode. That being said, since these putative coding regions have not been thoroughly characterised, one should be very careful about homology (similarity in sequence) based on direct ancestor descendent relationship (orthology) or based on being derived from a common ancestral sequence through gene duplication (paralogy). If you study organismal evolution / phylogenetics, then including paralogs can have very detrimental effect on the analysis. If the purpose is to study the evolution of a gene / gene family, then the inclusion of these putative gene sequences can shed light on duplication events. So it very much depend on your objective to what extent these sequences will turn out to be useful.

I am a bit worried by some of these answers, I must say !

I would rationalize my own approach in several critical steps:

1. Is the sequence you are talking about a nucleotide or a protein sequence ? Currently we are in the very vast majority of cases relying on nucleotide sequences as primary data. Protein sequences are in silico derived from it, by machines with or without the supervision of a human expert. In most cases, without ...

2. Is the sequence you are talking about the one from a transcript or a genomic one? In the first case you can use it as such, remembering that there are often many alternative transcripts for a single gene... And that pseudogenes are occasionally transcribed as well ... in which case such a transcript would be misleading you. And that transcripts happen to encode proteins, ... or not, quite many acting as RNA, overlapping protein-encoding genes in the antisense or sense orientation, with a sequences which are then quite far or completely unrelated to the one of the CDS of the overlapped gene.

3. If it is a genomic sequence, is it from a prokaryote or from a eukaryote ? In the first case there are big chances that the CDS & encoding protein will be valid, the most important problem being there to get the proper translation start, which can be tricky ...

3. If you are dealing with eukaryotes, then you are far from done with your DNA sequence! Someone may have been predicting a gene model from it, i.e. the CDS exon-intron structure, and this can can be correct or terribly wrong ...

4. Staying on what looks to be the safe side, nucleotides, you have to remember that such sequences are most often not the real primary data, which are reads... Meaning that afterwards a machine/someone has assembled these reads into the sequence you are dealing with, correctly or not. The worst being for comparative genomics the paralogous genes & transcripts and other sources of repeats which may end up in chimeric assembly.

All this being taken into account, it is quite often that you nevertheless can use unannotated or poorly annotated sequences for comparative purposes !

Good luck

Pierre

Thanks to Fernando for his remark on the common confusion between homology and similarity. Now as for the original question on whether or not hits from blast searches can be included in a phylogenetic study, the answer is yes without any restriction. Just check that the hit encodes a putative protein of the expected family.

I agree with previous replies. You can use the sequences. The sequences are not predicted (their status as genes might be, or their annotation, but that does not count for phylogeny). More generally, if the sequences turn out not to be homologues, you should see this clearly as a result of your phylogenetic analysis (they should cluster together as an outgroup). The only worry would be that they would "contaminate" your alignment by breaking the underlying homology hypothesis, but the fact that you can find them with BLASTn pretty much makes sure this is not the case.

Phylogenetic analysis and evolutionary studies should be performed using the software, cyber and information systems and laws. In this study should use new digital chart genetic code, to decode the codes in these sequences. Answers to all questions should be asked in the digital codes. These digital codes is easy to decipher. Digital genetics will, sooner or later, to replace the existing empirical genetics.

Yeah I agree with Brian -- the annotation "predicted" before annotating the function of a gene just means that the function is likely to be the one given based on the similarity between the sequence in question with the sequence of a protein whose function has been characterized. When reporting your functional assignment for this gene, you can likewise be conservative in your language and say that the function is "predicted." The reality we work in right now is that most proteins out there have not yet been isolated and characterized biochemically. Unless you're making an extraordinary claim about the function of this gene, most folks would give the benefit of the doubt that high sequence similarity implies homology and, thus, probably the same or similar function. The functional assignment would be especially well-supported if the rest of the pathway involving this function is present in your experimental genome. However, these predictions *can* be wrong, and have been! (One example is the predicted formate dehydrogenase in /Dehalococcoides ethenogenes/, which is *not* a formate dehydrogenase.) To further support your functional evidence, you could align the protein sequence from your gene with the homologous annotated proteins on NCBI, and then look through the annotations of the NCBI genes for where the active site residues are. If your gene is much more conserved with the others within the critical functional sites, that could provide further support that they are all likely to share the same function.

So, yes, go ahead and use those sequences!

Then, there's the question of using these sequences to investigate evolutionary history (phylogeny, or in a simpler sense, taxonomy) - For some genes this works very well (e.g. methyl CoM reductase in archaea is quite variable and fairly well-resolved in terms of phylogeny), but for other genes you can't resolve evolutionary history at all (e.g., if you're looking at an acetyl CoA synthetase sequence, you can't even tell what domain it came from, since it is so well conserved). You could try building a phylogenetic tree from the collection of known sequences and see how well it resolves taxonomy.

I agree almost with all answers !

The questions you need to answer in order to get a good answer are:

- what is you purpose for running a phylogenetic analysis ?

- what is the origin of your sequence(s), i.e. do you already know the species ? if yes is it a model species ? is it derived from an environmental sample ?

The first thing to do is to determine if this is a protein coding sequence or not. Blast will not tell you that ! (as for exemple in the case of a pseudogene)

If yes:

- Blast at Uniprot, since annotation are by far much better than at genbank.

- In case the answer is negative or similarity bad, go for functional domains search, in order to have an idea about the function of your protein

... other suggestion depend on your answers to questions above.

You can use predicted sequences for phylogenetic analysis and evolutionary studies. In this respect it should do the following: Each sequence set a numeric value. Let it be the number of atoms. When done, I will suggest what else to do.

@ Lutov Kuric, the point is not much clear to me.

How can I set a numeric velue to a sequence, based on the number of atom? How is it correlated with the acceptability of predicted sequence in phylogeny based on molecular features?

Please explain.

You can, and very successfully. The mathematical analysis is relatively simple. First you should, instead of chemical symbols sequences, put the numerical value of each sequence. That way you'll biochemistry into mathematics. Then you will, with a little effort to uncover the corresponding sequences, and which are not. In fact, you will receive a mathematical answer to your question. After you answer the language of mathematics to translate into the language of genetics. It may seem complicated, but it is not ...

@ Lutvo Kuric, thanks. It is entirely a new way of analysing molecular phylogeny, to me.

Can you give a very small example, and show it can be done.

Can. Need to send me a fragment sequences in this fragment should specify the following information: 1) sequences (amino acids or nucleotides), 2) the number of atoms in each sequence.

If possible, give me a fragment or fragments that represent a whole. After that I will propose to you how to proceed with this research.

@ Lutov, I am sending you the sequence shortly. I need to get some time to select a gene sequence (small in size) with amino acid sequence. This week-end I am pre-occupied with some out-station work.

Thanks.

@ lutov, i have just selected 5 sequences (nucleotide and translated amino acids):

however, i could not determine the number of atom, please tel me how to determine that one, both for dna and protein sequences.

The sequences are as follows:

> nm_001014385bostaurus

atgttcagagtattgaacgacgatgtttacagtccagctgtggtacagcaacaaactactctcgctttta

ggaaagactcttccttgtgcacagacagtcatagcgcaaatgatcagtgtgaaaggggagaacatgttag

ggagagcagccaggaccacgtcaagcgacccatgaacgccttcattgtgtggtctcgtgaacgaagacga

aaggtggctctagagaatcccaaaatgaaaaactcagacatcagcaagcagctgggatatgagtggaaaa

ggcttacagatgctgaaaagcgcccattctttgaggaggcacagagactactagccatacaccgagacaa

atacccgggctataaatatcgacctcgtcggagagccaagaggccacagaaatcgcttcctgcagactct

tcaatactatgcaacccgatgcatgtagagacattgcaccccttcacatacagggatggttgtgccaaga

ccacatactcacaaatggaaagccaattaagccggtcacagtccgtgatcataaccaattcactcctgca

aaaggagcatcacagcagctggacaagcctgggccacaataaggtaacattggctacacggatttcggcg

gactttccctgtaacaaaagcttagagcctggactttcttgtgcttattttcaatattga

> dq336531bosgrunniens

atgttcagagtat

tgaacgacgatgtttacagtccagctgtggtacagcaacaaactactctcgcttttaggaaagactcttc

cttgtgcacagacagtcatagcgcaaatgatcagtgtgaaaggggagaacatgttagggagagcagccag

gaccacgtcaagcgacccatgaacgccttcattgtgtggtctcgtgaacgaagacgaaaggtggctctag

agaatcccaaaatgaaaaactcagacatcagcaagcagctgggatatgagtggaaaaggcttacagatgc

tgaaaagcgcccattctttgaggaggcacagagactactagccatacaccgagacaaatacccgggctat

aaatatcgacctcgtcggagagccaagaggccacagaaatcgcttcctgcagactcttcaatactatgca

acccgatgcatgtagagacattgcaccccttcacatacagggatggttgtgccaagaccacatactcaca

aatggaaagccaattaagccggtcacagtccgtgatcataaccaattcactcctgcaaaaggagcatcac

agcagctggacaagcctgggccacaataaggtaacattggctacacggatttcggcggactttcccttta

acaaaagcttagagcctggactttcttgtgcttattttcaatattga

> dq336530bosfrontalis

atgttcagagtat

tgaacgacgatgtttacagtccagctgtggtacagcaacaaactactctcgcttttaggaaagactcttc

cttgtgcacagacagtcatagcgcaaatgatcagtgtgaaaggggagaacatgttagggagagcagccag

gaccacgtcaagcgacccatgaacgccttcattgtgtggtctcgtgaacgaagacgaaaggtggctctag

agaatcccaaaatgaaaaactcagacatcagcaagcagctgggatatgagtggaaaaggcttacagatgc

tgaaaagcgcccattctttgaggaggcacagagactactagccatacaccgagacaaatacccgggctat

aaatatcgacctcgtcggagagccaagaggccacagaaatcgcttcctgcagactcttcaatactatgca

acccgatgcatgtagagacattgcaccccttcacatacagggatggttgtgccaagaccacatactcaca

aatggaaagccaattaagccggtcacagtccgtgatcataaccaattcactcctgcaaaaggagcatcac

agcagctggacaagcctgggccacaataaggtaacattggctacacggatttcggcggactttcccttta

acaaaagcttagagcctggactttcttgtgcttattttcaatattga

> eu581862caprahircus

atgaatagaacggtgcaatcgtatgcttctgctatgttcagag

tattgaaagacgatgtttacagtccagcggtggtacagcaacaaaatactttcgcctttgggaaaacctc

ttccttgtgcacagacaatcatagtgcaaatgatcagtgtgaaaggggcgaaaatgttacggagagcagc

caggaccacgtcaagcgacccatgaacgccttcattgtgtggtctcgtgaacgaagacgaaaggtggctc

tagagaatcccaaattgcaaaactcagagatcagcaagcagctgggatacgagtggaaaaggcttacaga

tgctgaaaagcgcccattctttgaggaggcacagagactactagctatacaccgagacaaatacccgggc

tataaatatcgacctcgtcggaaagccaagaggccacagaaatcgcttgatgcagactctccaatactat

gcaaccagatggatgtagagacattgcaccccttcacatacagggacgattgtgccaagaccacacactc

acaaatggaaagccaattatgccgctcacagtccctgattctaaccaattcacttctgcaaaaggagcat

cacagcagctggacaaacctgggccacgatagggtaacattggatacacggatttccgcggactttccct

tttaccaaagcttagagcctgggctttcttgcgcttatgttcaatactga

> nm_003140homosapiens

atgcaatcatatgcttctgctatgttaagcgtattcaacagcgatgattacagtccagctgtgcaagagaatattcccgctctccggagaagctcttccttcctttgcactgaaagctgtaactctaagtatcagtgtgaaacgggagaaaacagtaaaggcaacgtccaggatagagtgaagcgacccatgaacgcattcatcgtgtggtctcgcgatcagaggcgcaagatggctctagagaatcccagaatgcgaaactcagagatcagcaagcagctgggataccagtggaaaatgcttactgaagccgaaaaatggccattcttccaggaggcacagaaattacaggccatgcacagagagaaatacccgaattataagtatcgacctcgtcggaaggcgaagatgctgccgaagaattgcagtttgcttcccgcagatcccgcttcggtactctgcagcgaagtgcaactggacaacaggttgtacagggatgactgtacgaaagccacacactcaagaatggagcaccagctaggccacttaccgcccatcaacgcagccagctcaccgcagcaacgggaccgctacagccactggacaaagctgtag

=========================================

amino acid sequence

> nm_001014385bostaurus

mfrvlnddvyspavvqqqttlafrkdsslctdshsandqcergehvressqdhvkrpmnafivwsrerrrkvalenpkmknsdiskqlgyewkrltdaekrpffeeaqrllaihrdkypgykyrprrrakrpqkslpadssilcnpmhvetlhpftyrdgcakttysqmesqlsrsqsviitnsllqkehhsswtslghnkvtlatrisadfpcnkslepglscayfqy

> dq336531bosgrunniens

mfrvlnddvyspavvqqqttlafrkdsslctdshsandqcergehvressqdhvkrpmnafivwsrerrrkvalenpkmknsdiskqlgyewkrltdaekrpffeeaqrllaihrdkypgykyrprrrakrpqkslpadssilcnpmhvetlhpftyrdgcakttysqmesqlsrsqsviitnsllqkehhsswtslghnkvtlatrisadfpfnkslepglscayfqy

> dq336530bosfrontalis

mfrvlnddvyspavvqqqttlafrkdsslctdshsandqcergehvressqdhvkrpmnafivwsrerrrkvalenpkmknsdiskqlgyewkrltdaekrpffeeaqrllaihrdkypgykyrprrrakrpqkslpadssilcnpmhvetlhpftyrdgcakttysqmesqlsrsqsviitnsllqkehhsswtslghnkvtlatrisadfpfnkslepglscayfqy

> eu581862caprahircus

mnrtvqsyasamfrvlkddvyspavvqqqntfafgktsslctdnhsandqcergenvtessqdhvkrpmnafivwsrerrrkvalenpklqnseiskqlgyewkrltdaekrpffeeaqrllaihrdkypgykyrprrkakrpqksldadspilcnqmdvetlhpftyrddcaktthsqmesqlcrsqsliltnsllqkehhsswtnlghdrvtldtrisadfpfyqslepglscayvqy

> nm_003140homosapiens

mqsyasamlsvfnsddyspavqenipalrrsssflctescnskyqcetgenskgnvqdrvkrpmnafivwsrdqrrkmalenprmrnseiskqlgyqwkmlteaekwpffqeaqklqamhrekypnykyrprrkakmlpkncsllpadpasvlcsevqldnrlyrddctkathsrmehqlghlppinaasspqqrdryshwtkl

================================================================

Thank you. These sequences are very suitable for decoding and to discover the many secrets.

However, I have one technical problem. I do not have a program to translate these nucleotides with the language of genetics into the language of mathematics.

The first step in this research is to express sequences in this way

Nucleotides t t g g c

Number of atoms 15 15 16 16 13

Position 1 2 3 4 5

Etc.

After that we need to decode the codes that are hidden in these sequences.

This can be done with the following conditions:

Yes we send these sequences in the form of the previous table

-It is a voluminous work. I'm ready to do it with financial support.

Thanks!

Lutov Thanks a lot.

I will do the mathematical evaluation of sequence, as shown by you.

Please let me know the atom no. of the neucleotides and amino acids.

Thanks.

I suggest you take the numerical values ​​of nucleotides> nm_001014385bostaurus And then decodes these sequences. When I do this I will send you the results derived. If you are satisfied with these results, I will then decode and all other sequences. So all I will do. I will do this if we accept that after decoding sequences nm_001014385bostaurus given financial support. I expect similar support after completion of the entire study. The methods used in the study are very reliable and give accurate results stoposto. That job can be completed in a relatively short time. All copyrights will belong to you. If you accept that, let me know.

Thanks;

What is the required amount of Financial Support?

I suggest to you that you decide on. I do not know how to decode the complex. I am a humble man and I have great demands. When you review the first sequence to the judge. I think we can easily agree.

Thanks!

Dear Lutov Kuric, thanks for your extended help.

I have cogitated over the matter and I think that our discussion should involve constructive discussion only, instead of financial deals.

Best

'Predicted' genes are exactly as they are described: only predicted. They may or may not be real.

Click on the details of the gene and the level of evidence will be documented. Click all the links and read it all. Usually, the prediction is based on computational analysis of nucleotides showing similarity to a known protein. 

What is a predicted sequence in NCBI? - ResearchGate. Available from: https://www.researchgate.net/post/What_is_a_predicted_sequence_in_NCBI [accessed Dec 13, 2016].