Two seqences are being shown to be 100% similar in BLAST search but performing pairwise alignment via the EMBL server is showing the sequences to be around 46% similar. Why this difference in similarity is being shown?
Blast gives identity based in local alignment and pairwise alignment which you might be performing is giving you global similarity information. Check your complete blast results before proceeding further.
The two use different algorithms. Blast is designed to be fast, as it is used to compare a query sequence to large sequence databases and identify those that show significant similarity. https://en.wikipedia.org/wiki/BLAST_(biotechnology)
For pairwise sequence alignment, you can afford to use less efficient, but more precise algorithms.
The value of % similarity between two sequences depends on the placement of alignment gaps, which in turn depends on gap insertion and gap extension penalties used, on the scoring matrices used to define the similarity between to different amino acids in the same position, and on whether the 100% value references the length of the longer sequence, the shorter sequence or the fraction of the sequence that could be successfully aligned. It is not a value which is defined independent on the algorithm and parameters used to generate an alignment. Even if used for a qualitative comparison, you have always to check whether it refers to the same number of residues: one program might tell you that you have a very high similarity over a short stretch of the sequence, while the other might include a larger fraction of the sequence.
Reverse complement one of the sequences and alighn using the same
Check the query cover of the sequences. BLAST may only align a segment of the two sequences (the regions with good coverage) and ignore other regions with poor/no homology. Hence, the high sequence identity reported by BLAST may not
be for the entire sequence length but for only the smaller local aligned region.
- Badges
- Science method