blast.ncbi.nlm.nih.gov

BLAST (Basic Local Alignment Search Tool) finds regions of local similarity between biological sequences. It compares a nucleotide or protein query against sequence databases and calculates the statistical significance of matches, which lets you infer functional and evolutionary relationships between sequences and identify members of gene families. Use the standard Web BLAST when you have one sequence and want to know what it resembles; switch to a specialized search when your goal is narrower, such as designing primers, comparing two sequences end to end, or locating conserved domains.

What BLAST actually does

BLAST does not align two sequences from start to finish. It looks for local regions of similarity — short stretches where your query and a database sequence match well — and then scores those regions.

Two outputs matter most when you read a result:

• Alignment score — how well a local region matches; higher means a better match under the scoring scheme used.
• E-value — the expected number of matches this good that would occur by chance in a database of that size. Smaller is stronger.

Because matches are local, a single query can produce several separate hits, and a hit can cover only part of your sequence. That is a feature, not an error: it is how BLAST detects shared domains between otherwise different proteins.

Choosing the right search

The BLAST site groups its tools by purpose. Match the tool to your goal:

If you are unsure, start with Web BLAST. The specialized tools exist because the general search answers a different question — Primer-BLAST, for example, is built around primer specificity to a template, not around finding the closest relative of a sequence.

Running BLAST beyond the website

The web interface is one of several ways to run BLAST. The site also offers:

• Download BLAST — get BLAST databases and executables to run locally.
• Use BLAST API — call BLAST from your own application.
• Use BLAST in the cloud — start an instance at a cloud provider.

The choice is about scale and automation: interactive one-off queries fit the web interface, while repeated or programmatic searches fit the API, local executables, or a cloud instance. Local and cloud options also matter when you need to search custom databases or control where your data lives.

Reading results without over-reading them

A strong hit is evidence, not proof. Before treating a match as definitive:

• Check coverage — does the alignment span most of your query, or just a fragment?
• Check the E-value in context of the database size, not as an absolute pass/fail.
• Expect multiple hits for multi-domain proteins; each may reflect a different shared region.
• Treat functional or evolutionary inference as a hypothesis to confirm by other means.

The practical rule: use BLAST to generate candidates and relationships, then verify the ones that matter.

Sequence homology is a conclusion about shared ancestry, not a measurement. Two sequences are homologous if they descend from a common ancestral sequence; you cannot assign a "percent homology." What BLAST gives you is sequence similarity — local regions where a query and a database sequence align well — plus statistics (E-value, bit score) that tell you how unlikely that similarity is to arise by chance. You then use that evidence, together with alignment coverage and biological context, to decide whether a homology claim is defensible. This applies to nucleotide and protein searches against public databases via NCBI BLAST.

Similarity vs. homology: keep the two separate

• Similarity is observable and quantitative: percent identity, alignment length, gaps, bit score.
• Homology is an inference about history: either the sequences share an ancestor or they do not. There is no such thing as "40% homologous."
• High similarity is evidence for homology; it is not the same statement. Two short, low-complexity regions can be highly similar by chance, and two true homologs can diverge until similarity is faint.

BLAST's own description frames this correctly: it "finds regions of local similarity between sequences" and "can be used to infer functional and evolutionary relationships between sequences as well as help identify members of gene families." The inference step is yours.

What BLAST actually computes

BLAST compares your query against a sequence database and reports local alignments — the best-matching subregions, not a forced end-to-end comparison. For each hit you get:

The E-value is the field most often misread. It is not a probability that the match is homologous, and it is not a p-value in the usual sense — it is an expectation over the whole database search.

Reading a BLAST result to judge homology

Work through the evidence rather than stopping at the top hit:

1. Check the E-value in context. A very small E-value (for example, far below 1e-5 for a typical protein search) is strong evidence the similarity is not random. Borderline values need corroboration.
2. Check coverage. If the alignment covers only a small fraction of your query, a high percent identity over that fragment may reflect a shared domain rather than full-length homology.
3. Look at the alignment itself. Are the matches spread across the length, or concentrated in a repetitive or low-complexity stretch? BLAST masks low-complexity regions by default for good reason.
4. Compare multiple hits. If several unrelated database sequences all align to the same region of your query with similar scores, suspect a common domain or a compositional bias rather than orthology.
5. Bring in outside evidence. Conserved domain searches, synteny, or experimental annotation strengthen or weaken the inference. BLAST alone rarely settles a homology question.

Common pitfalls

• Equating high identity with homology. A 95% identity over 30 residues is weak evidence; 30% identity over 300 residues with good coverage may be strong.
• Ignoring database size. The same alignment yields a worse E-value in a larger database. A hit that looks significant in a small custom database may not be in a comprehensive one.
• Treating the top hit as the true homolog. The top hit is the best-scoring match in that database at that moment, which is not the same as the closest evolutionary relative.
• Forgetting that "homolog" needs a qualifier. Once you accept homology, specify whether you mean orthologs (speciation) or paralogs (duplication), because the functional implications differ.
• Over-reading percent identity across species. Rapidly evolving or saturated sequences can hide true homology; profile- or structure-based methods may recover relationships that pairwise BLAST misses.

What you can do once homology is supported

• Function annotation: transfer a known function from a characterized homolog to an uncharacterized sequence, with the caveat that function does not always track ancestry.
• Gene family assignment: group sequences into families using consistent thresholds and coverage rules, and document them.
• Evolutionary analysis: build alignments and trees from sequences you have justified as homologous — the homology call is the prerequisite, not the output, of phylogenetic analysis.

The practical rule: report similarity as measured, and state homology as an inference with the evidence and thresholds you used.

Sequence similarity is the degree to which two biological sequences — DNA, RNA, or protein — match in local regions when aligned side by side. BLAST (Basic Local Alignment Search Tool) measures it by comparing your query sequence against a sequence database, finding regions of local similarity, and calculating the statistical significance of those matches. You use BLAST when you have a sequence and want to know what it resembles, whether that is to infer function, trace evolutionary relationships, or identify which gene family a sequence belongs to.

Similarity, identity, and homology are not the same thing

These three terms get used interchangeably in casual conversation, but they answer different questions:

Similarity and identity are measurements. Homology is an inference you draw from those measurements. A high similarity score is evidence for homology, but it is not proof on its own — significance testing is what turns a raw score into a defensible claim.

How BLAST finds and scores local alignments

BLAST does not try to align your entire query against an entire database entry. It searches for local regions where the sequences match well, which is why it can detect a shared domain even when the surrounding sequence is unrelated.

The process works roughly like this:

1. Seeding. BLAST breaks the query into short words and looks for database words that score above a threshold. These seed hits are candidate alignment starting points.
2. Extension. Each seed is extended in both directions, adding matches and penalizing mismatches and gaps according to a scoring scheme (for example, a substitution matrix plus gap penalties for proteins).
3. High-scoring segment pairs (HSPs). The extension stops when the score stops improving. What remains is an HSP — a local alignment with a raw score.
4. Statistical evaluation. The raw score is converted into an E-value, which is the number of matches with that score or better you would expect to find by chance in a database of that size.

The E-value is the number that matters most for interpretation. It depends on both the alignment score and the size of the database searched, so the same alignment can produce different E-values against different databases.

Reading the E-value

The E-value answers: how often would a match this good appear by random chance?

• Very small E-values (for example, below 1e-5, and often far smaller) indicate the match is unlikely to be a random artifact. This is the range where functional and evolutionary inferences become reasonable.
• E-values near or above 1 mean you would expect matches this good by chance, so the hit carries little information.
• E-value thresholds are conventions, not laws. A common cutoff is 1e-5, but the right threshold depends on your database size, sequence length, and how much tolerance you have for false positives.

A low E-value is necessary but not sufficient for a biological conclusion. A short query, a low-complexity region, or a biased composition can all produce misleadingly good scores.

What a significant match lets you infer

When a match is statistically significant, BLAST's own documentation notes that the tool can be used to infer functional and evolutionary relationships between sequences and to help identify members of gene families. In practice:

• Function transfer. If your unknown protein aligns strongly to a characterized protein, the shared similarity suggests related function — but only as a hypothesis to test, especially if the match covers only part of the sequence.
• Evolutionary relationship. Strong similarity across a conserved region is consistent with shared ancestry. The strength and extent of the alignment shape how confident that inference is.
• Gene family membership. A significant hit to a known family member is often how a new sequence gets assigned to a family.

The key caution: significant similarity supports these inferences, it does not establish them. Domain-level matches, partial alignments, and matches to uncharacterized entries all require judgment beyond the E-value.

Practical checks before you trust a hit

• Look at alignment coverage, not just the score. A high-scoring match to a small fragment of your query tells you less than a moderate match across the full length.
• Check the database. Matches to well-annotated reference sequences carry more interpretive weight than matches to uncharacterized entries.
• Watch for low-complexity or repetitive regions, which can generate spurious high scores.
• Compare against more than one hit. A single strong hit is informative; a cluster of hits to related sequences is stronger evidence.

BLAST's specialized tools — SmartBLAST for finding highly similar proteins, Primer-BLAST for primer design, Global Align for full-length comparison, and CD-search for conserved domains — each apply similarity measurement to a narrower question, so the interpretation guidance above applies with the appropriate adjustments for each.

Sequence alignment is the process of arranging two or more biological sequences—DNA, RNA, or protein—so that positions with shared ancestry or similar function line up with each other. BLAST (Basic Local Alignment Search Tool) performs a specific kind of alignment: it finds short, local regions of similarity between a query sequence and sequences in a database, then calculates the statistical significance of each match. You would use BLAST when you have a sequence and want to know what it resembles, rather than when you need a full end-to-end alignment of two sequences you already know are related.

What sequence alignment actually produces

An alignment is a set of positional correspondences. When two sequences are aligned, matching positions are placed in the same column, and gaps are inserted where one sequence has residues the other lacks. The output is a row of paired characters plus gap symbols, which lets you count identities, similarities, and differences position by position.

Two sequences can be aligned in fundamentally different scopes:

• Local alignment finds the highest-scoring region of similarity within otherwise dissimilar sequences. It does not require the rest of the sequences to match.
• Global alignment forces the two sequences to be compared across their entire span, from start to end, penalizing gaps and mismatches everywhere.

BLAST is a local alignment tool. That design choice is why it can detect a shared protein domain in two sequences that are otherwise unrelated, and why it does not try to stretch a match across regions that have no similarity.

How BLAST finds local regions of similarity

BLAST does not attempt an exhaustive alignment. It uses a seed-and-extend strategy that trades some sensitivity for large gains in speed, which is what makes database-scale searching practical.

The general mechanism:

1. Word generation. The query sequence is broken into short words (for protein searches, a fixed small number of residues). BLAST compiles a list of words that score above a threshold against each query word.
2. Database scan. The database is scanned for exact or near-exact hits to those words. These are the seeds.
3. Extension. Each seed is extended in both directions, adding positions as long as the alignment score keeps improving. When the score stops rising, the extension halts.
4. Reporting. High-scoring segment pairs (HSPs) that survive a score cutoff are reported as local alignments.

The result is a set of local alignments, each covering only the region where the two sequences genuinely resemble each other.

How BLAST calculates statistical significance

A raw alignment score depends on the scoring matrix, gap penalties, and sequence length, so it is not directly comparable across searches. BLAST converts the raw score into an E-value (expectation value), which estimates how many alignments with a score at least that high would be expected by chance alone in a database of that size.

The practical reading of an E-value:

Because the E-value scales with database size, the same alignment can yield a different E-value in a larger database. The site describes BLAST as comparing sequences to sequence databases and calculating the statistical significance of matches, which is exactly this step.

Local vs. global alignment: choosing the right tool

The distinction matters when you pick a method. BLAST's own specialized search list makes the contrast concrete: Global Align compares two sequences across their entire span using the Needleman-Wunsch algorithm, while BLAST itself reports local regions.

Choose BLAST when you want to discover similarity you did not know was there. Choose a global alignment when you have two sequences of comparable length and want a complete positional comparison.

What alignment results tell you

A significant local alignment is evidence of homology—shared evolutionary origin—which in turn supports inferences about function. If an uncharacterized protein aligns strongly across a conserved domain to a protein with a known function, that domain is likely present and performing a related role. BLAST's stated uses reflect this: inferring functional and evolutionary relationships between sequences and identifying members of gene families.

Two cautions when reading results:

• Similarity is not identity of function. A strong alignment indicates relatedness; it does not by itself prove the two sequences do the same thing in the cell.
• Local matches can be partial. A high-scoring HSP may cover only one domain of a multi-domain protein, so the overall proteins may differ substantially outside that region.

Getting started

BLAST is available through several entry points described on the site:

• Web BLAST — run searches in the browser, including genome searches by organism name or tax ID.
• Standalone and API BLAST — download executables and databases, call BLAST from an application, or run it in the cloud.
• Specialized searches — SmartBLAST for finding highly similar proteins, Primer-BLAST for designing PCR primers, CD-search for conserved domains, IgBLAST for immunoglobulin and T cell receptor sequences, VecScreen for vector contamination, and CDART for conserved domain architecture.

For a first local alignment, start with Web BLAST: paste your nucleotide or protein query, select the appropriate program and database, run the search, and read the reported HSPs alongside their E-values. If instead you need a full end-to-end comparison of two sequences, use Global Align rather than BLAST.

Bioinformatics is the practice of applying computation, statistics, and software to biological data — DNA, RNA, and protein sequences, plus the measurements derived from them. You'd consider studying or working in it if you like biology but are equally comfortable with code, data, and ambiguity, and you want a role where the deliverable is often a working pipeline or a defensible result rather than a wet-lab experiment. The field overlaps heavily with computational biology and genomics, but the three are not identical, and knowing the difference helps you read job posts accurately.

Where the boundaries sit

In practice the labels blur. A genomics group hires bioinformaticians to build the analysis; a computational biology lab may write its own tools. When you read a job description, look at the verbs: "build," "maintain," and "process" point toward bioinformatics; "model," "simulate," and "hypothesize" point toward computational biology.

What the day-to-day work actually looks like

Most of the job is not glamorous discovery. A realistic week includes:

• Data cleaning and quality control. Raw sequencing reads arrive with adapters, low-quality bases, and contamination. You trim, filter, and check quality metrics before anything else.
• Sequence alignment and mapping. You align reads to a reference genome or compare sequences against each other, then interpret the alignment statistics.
• Variant calling and annotation. You identify differences from a reference — single-nucleotide variants, insertions, deletions — and attach biological meaning to them.
• Pipeline building. You wire tools together so the same analysis runs reproducibly on new samples, often with workflow managers rather than by hand.
• Visualization and reporting. You turn tables and alignments into figures and summaries that a biologist, clinician, or collaborator can act on.

A large share of time goes to the first and last items. Anyone entering the field should expect debugging and documentation to be a core part of the work, not an afterthought.

Core skills and tools

• Programming. Python and R are the two dominant languages. Python tends to dominate pipeline and tool development; R dominates statistical analysis and visualization.
• Command line. Most bioinformatics tools run in a Unix-like shell, so comfort with the terminal is effectively mandatory.
• Statistics. You need to understand what a p-value, a false discovery rate, and a batch effect mean, because biological data is noisy and confounded.
• Databases and formats. You'll work constantly with public resources such as NCBI and UniProt, and with file formats like FASTA, FASTQ, SAM/BAM, and VCF.
• Reproducibility practices. Version control, environments, and workflow managers separate a one-off script from something a lab can rely on.

Where people work

• Academia and research institutes — core facilities, labs, and consortia, often with a mix of service work and independent projects.
• Biotech and pharma — drug discovery, target identification, clinical trial data, and increasingly large-scale production pipelines.
• Clinical labs — variant interpretation and reporting, where accuracy and regulatory constraints are stricter than in research.
• Software and data companies — building the tools and platforms other scientists use.

Roles range from research scientist to bioinformatics engineer to data scientist, and the split between "build the tool" and "use the tool" varies by employer.

Getting started and staying current

Bioinformatics.org describes itself as a community open to all people, with a strong emphasis on open access to biological information and free and open source software. It lists education and courses, jobs and careers, and news among its sections, and its featured prior sponsors include Regeneron, St. Jude, Illumina, BioSpace, Georgetown University, and Northeastern University. The site also notes a subscribe option for news, so if you want updates, check the current terms on the site rather than assuming access is unrestricted.

A practical path for a beginner: pick one public dataset, run a standard alignment and variant-calling workflow end to end, and write up what you found and where it broke. That single exercise touches nearly every skill listed above and gives you something concrete to discuss in an interview or application.