Understanding the concepts of blast score and E-value is crucial in bioinformatics, especially when dealing with sequence alignments and databases. These metrics help us determine the statistical significance of our results, and thus, the confidence we can place in them. Let's dive right in to understand these concepts in detail.

Sequence alignments allow us to compare DNA, RNA, or protein sequences and identify similarities. However, with vast databases containing millions of sequences, it's essential to have a method to judge the significance of our results. This is where blast score and E-value come into play.

Blast Score
The blast score, often referred to as the raw score, is a simple score assigned for each alignment. It corresponds to the number of identical matches between the query and subject sequences, adjusted according to the size of the gap between aligned residues.

However, blast scores alone can be misleading. Two alignments with the same blast score may have quite different levels of statistical significance. This is where the concept of E-value comes in, providing a more useful measure of the random chance of obtaining a similar score by chance.
Calculating Blast Score

The blast score is calculated using the following formula: S = Σ(s[i] * M[i] / G), where:
- S is the raw score
- s[i] is the similarity score for each match
- M[i] is the cost of the mismatch or gap
- G is the maximum of the score for a mismatch and the cost of opening a gap
The blast program then adds up these individual scores for all matched positions to get the final blast score.

Interpreting Blast Score
The blast score doesn't tell us everything about an alignment. It's the raw score, and without additional information, we can't tell if the alignment is statistically significant. For that, we need to consider the E-value.
For instance, a blast score of 50 for two sequences could be significant or not, depending on their length. If the sequences are short, a blast score of 50 might indicate a high degree of similarity. However, if the sequences are longer, this score might not be very impressive.

E-value
The Expect (E) value in blast gives the probability of observing the alignment score by chance. It estimates the number of alignments with a given score that we would expect to see by chance alone, assuming the sequences have no true homologies.








E-values are usually expressed as a probability, where smaller values indicate better significance. By comparing the E-value to a chosen threshold (usually 0.01-0.05), we can determine whether our alignment is significant or not.
Calculating E-value
The calculation of the E-value is based on the blast score (S) and the length of both the query sequence (l_query) and the database sequence (l_database). It uses the following formula:
E = l_query * l_database * k * e^(-S/λ)
where:
- k is the substitution rate (usually 0.2 for protein sequences)
- λ is the average match score (usually 2.0 for protein sequences)
The exact values of k and λ can be adjusted depending on the specific blast program and the sequences being compared.
Interpreting E-value
An E-value of 1 means that in a database search of the size conducted, one match is expected to occur by chance alone. Therefore, an E-value of 0.01, or 0.05, means that only one, or five, matches are expected by chance, respectively.
As the E-value gets smaller, the significance increases, indicating that the match is less likely to occur by chance alone. Thus, alignments with small E-values are generally considered significant, whereas those with large E-values are not.
Applying E-value Thresholds
Setting an appropriate E-value threshold is crucial. A reasonable E-value cutoff is typically in the range of 0.01 to 0.05, depending on the analysis and its purpose. For example, a more stringent E-value threshold might be used when searching for highly similar sequences in a well-studied organism, while a less stringent E-value might be used when searching for distant relatives.
Remember, setting a high threshold may lead to many false positives, while setting a low threshold may result in many false negatives. It's about finding the right balance between sensitivity and specificity.
Understanding and applying blast scores and E-values allows us to draw accurate conclusions from our sequence alignments. It takes us from simply seeing that two sequences share some similarity to being able to quantify that similarity and assess its statistical significance. This is paramount in the vast and complex databases we navigate in bioinformatics.