BLAST Optimization for Better Sequence Search Results

The Basic Local Alignment Search Tool, or BLAST, is a program that allows researchers to compare a DNA or protein sequence to vast libraries of known sequences. This process is analogous to a search engine, but the query is a biological sequence instead of words. Its applications are widespread in biology and medicine, from identifying an unknown species based on its DNA to discovering homologous genes across different organisms. Understanding how to refine a BLAST search is a valuable skill for obtaining clear and meaningful results.

Preparing Your Input Sequence

A BLAST search begins with a query sequence, which is the DNA or protein sequence a researcher wants to investigate. The quality of this input directly influences the quality of the results. Before a search, ensure the sequence is clean and correctly formatted. The most common format is FASTA, a text-based system where a description line, starting with a “>” symbol, is followed by the raw sequence data.

Problems in the query sequence can lead to misleading results. A common issue is vector contamination, where DNA used in laboratory sequencing procedures remains attached to the sequence of interest. These vector sequences are not part of the biological sample and can generate strong, but irrelevant, matches. Low-quality data, such as ambiguous bases (‘N’), can also weaken the potential for finding a legitimate match. Trimming vectors and cleaning up ambiguous regions is a foundational step for an effective search.

Choosing the Correct BLAST Program and Database

The term “BLAST” refers to a suite of programs, each designed for a specific type of comparison. Selecting the right program is a primary step, and the choice depends on whether the query is a nucleotide or protein sequence. The main programs include:

• BLASTN: Compares a nucleotide query against a nucleotide database.
• BLASTP: Compares a protein query against a protein database to find similar proteins.
• BLASTX: Translates a nucleotide query in all six reading frames and compares the results against a protein database, which is useful for finding protein-coding genes.
• TBLASTN: Compares a protein query against a nucleotide database that is dynamically translated in all six frames.
• TBLASTX: Compares the six-frame translation of a nucleotide query against the six-frame translation of a nucleotide database.

Matching the program to a suitable database is also necessary. Databases range from the comprehensive but less curated to the specific and highly annotated. Common choices include:

• The ‘nr’ (non-redundant) protein and ‘nt’ nucleotide databases, which are vast repositories from many sources.
• The Reference Sequence (RefSeq) database, which offers a curated, non-redundant set of genomic, transcript, and protein sequences.
• The Swiss-Prot database, which provides high-quality, manually annotated protein records.
• The Protein Data Bank (PDB), which contains sequences derived from experimentally determined 3D structures.

Adjusting Key Search Settings

Several search parameters can be adjusted to refine the results. One of the most significant is the Expect value, or E-value, threshold. The E-value describes the number of hits one can expect to see by chance when searching a database of a particular size. Setting a lower, more stringent E-value (e.g., 0.01) will filter out weaker matches, while a higher E-value might be used for exploratory searches.

Another parameter that affects search sensitivity is the word size. BLAST initiates alignments by looking for short, perfectly matching “words” between the query and database sequences. Reducing this size makes the search more sensitive and able to find more dissimilar sequences, but it also increases run time. Increasing the word size makes the search faster but may cause it to miss some alignments.

Filters can be applied to prevent certain parts of the query from skewing the results. Low-complexity regions are sequences with biased composition, such as long repeats of a single amino acid. These regions can produce high-scoring but biologically uninteresting alignments. Activating the low-complexity filter masks these regions, allowing the search to focus on finding more meaningful patterns of similarity.

Making Sense of BLAST Output

The results page presents several metrics to help interpret the findings. A primary indicator is the bit score, a normalized score that reflects the quality of the alignment; a higher bit score indicates a better match. This score is used to calculate the E-value for each hit.

Two other metrics are percent identity and query coverage. Percent identity states the percentage of characters that are identical between the query and the subject sequence in the aligned region. Query coverage shows what percentage of the input query’s length has aligned with the database sequence. High query coverage suggests the alignment spans nearly the entire query, while low coverage indicates only a small portion matched.

The output also provides a visual representation of the alignment, showing the query sequence stacked on top of the subject (database) sequence. This view allows for a direct inspection of where the sequences match, mismatches, and where gaps were introduced to optimize the alignment. Evaluating all these components together allows a researcher to form a comprehensive understanding of the results.