A phylogeny is only as good as the alignment it is built on.
When comparing generated sequence data with publicly available data (ie NCBI) there seems to be a variety of problems that can arise.
Even if one were to use identical primer sets as used to amplify available comparative data I envisage an issue.
The annealing sites of sequences at either end are commonly problematic with poor resolution, the only way to deal with such erroneous base pairs is to remove them?
Furthermore, sometimes, to ensure that you can compare data across a wide range of generated sequence data you may be required to reduce the alignment length down to have the largest comparable coverage. Especially as your alignment needs to have equal length for accurate phylogenetics analysis.
Surely, this alone is significant factor. If regions of high diversity happen to be those removed in the adjustment of the alignment then this could give a skewed representation of of genetic diversity and inaccurate phylogeny especially higher up the taxonomic stratification.
Would be interested to hear your comments, criticisms and general opinion on this issue and the above listed hurdles.
Cheers,
Hello Asif,
I totally agree with your first statement, the alignment of sequences is perhaps the most critical step in a phylogenetic analysis because the latter is based on the assumption that the sites are comparable, which is the objective of the multiple sequence alignment. Segments of sequences with poor resolution at the ends make that those sites be not comparable, so trimming is necessary and I don't see another way to deal with that, except increasing the length of the sequences by selecting another primers, or something analogous.
Regarding your second concern, you can reduce your alignment to some region that all your wider range of orthologies share, or conversely you can try extending the alignment by gapping the sequences that do not contain some region. In this case you avoid losing information that can be used by the phylogenetic algorithms to at least calculate distances between the sequences that share some region not present in the total of them. As a simple example just take a look at the global alignment of the SSU_rRNA gene (16S + 18S) in the Silva database from ARB (https://www.arb-silva.de/). All the sequences are extensively gapped in order to be comparable with the wide range of organisms considered (the whole tree of life!).
One can consider these analyses like a weather forecasting.. the wider the range the less precise are the results, as the model, by definition, is not the reality. The same happens in phylogenetics.
Best regards.
1) look first length - too much variation in length will be not recommended like one sequence is 1000 amino acid, while other sequences are 200 amino acid long, if DNA consider NT length
2) Sequences have to be clean without much XXXXXXXX, in case of DNA, NNNNNNNNNNN
3) Sequences must have some homology like u can fix two different superfamilies without prior knowledge why you are adding?
Thank you for your answer.
I guess this leads me to question the consistency of various people's methods and how that impacts upon perceived genetic diversity.
If researcher X's self generated sequence alingment is 600bp long while researcher Y's is 700bp long and to align them correctly gaps are inserted in X's alignment this would have a direct impact on downstream analysis.
The best way would be to ignore all gaps?
- Badges
- Science topic
- Similar topics
- Biological Science
- Molecular Biology
- Biomolecules
- Nucleic Acids
- DNA