Student Resources

BLAST, known as the Basic Local Alignment Search Tool, can identify similarities or differences within biological data. It accomplishes this by providing the user an E-value (Expected Value) to determine statistical significance. Explanations of each of the five types of BLAST and their uses can be found below:

Arrows indicate the BLAST program translates the nucleotide sequence before performing the search.

Each BLAST tool has a different function. You should understand that various tools of BLAST will provide you with different outcomes. If you use the wrong tool, you should expect information that doesn’t make sense and precious time lost. See the table below to see examples of how each tool of BLAST can be used in relation to your time with the GEP. If you ever see results that do not make sense, feel free to reach out to the GEP TA’s, and we will be more than happy to provide more in-depth explanations and assistance with your queries.

BLAST has three methods for the GEP to understand a target species and our reference species D. melanogaster. Instead of searching broad areas (all NCBI genomes) that may have no information pertaining to your query, BLAST can also be used for more narrow searches such as for Entrez and Assembly searches. It is easy to understand these different search queries as an overlapping funnel that varies based on what you are searching for with BLAST. The figures below show each query type and how they relate to the search results.

Reading the BLAST results page can be daunting at first. Be sure to anchor yourself to your initial scientific question so you don’t get lost. For a detailed breakdown of reading BLAST results, refer to the GEP Tools | NCBI BLAST video tutorial. Below is an example of a BLAST results page and a legend with specific explanations for how to interpret the page.

The RefSeq protein track found in the GEP UCSC Genome Browser is linked to RefSeq: NCBI Reference Sequence Database. When given a REFSEQ protein track, we need to ask the questions, “What is the predicted function of this gene?” and “Does this predictive function match what is found in D. melanogaster?” This, in turn, will give us a better scope of if we have located our target D. melanogaster homolog.

With a peptide sequence or even the Protein ID, we will need to use blastp. The tricky part of searching for a predictive function is that we don’t want to search for everything since this will take far too long and render many useless results. Consider the following questions:

• “What kind of input x and output f(x) am I expecting?”
  • Example: “I have a Protein sequence and am looking for Protein returns.”

• “Do I need results from hits outside of my genus, within the genus, or a specific species within my genus?”
  • Example: “I have a Protein sequence from D. kikkawai, and I am trying to find its predictive function from D. melanogaster.”

Assuming the answers to our questions above are “blastp” and “within the genus,” we can run this report by calling on an Entrez search with D. melanogaster (taxid: 2772).

BLAST relies on statistical analysis to determine areas between inputs and databases to produce statistically significant results. Sometimes when inputting a specific coding DNA sequence (CDS) or query, it is possible that BLAST will not warrant any results. This can happen for numerous reasons, three of the most common are listed below:

• No significant results are found. As simple as it sounds, sometimes, through enough evolution, CDSs can drastically change from one species to the next. If your species is several steps away from D. melanogaster, this CDS may be drastically different.
  • NOTE: When examining this kind of change, you can use the “Comparative Genomic” tabs in the GEP UCSC Genome Browser to provide evidence of this. Don’t expect to jump to this conclusion immediately – we are scientists and want to see why we are getting this result.
• The query submitted may not have been submitted correctly. Be sure to look back at which BLAST tool you are using and exam the query sequence you are comparing to it.
• Sometimes the information you are searching for can be “lost in translation.” This can cause problems with several search parameters, and BLAST ends up providing too little or too much and simply doesn’t want to flood you with information. To fix this kind of error, we suggest refining your search. A further explanation of refining your CDS can be found in the GEP | Pathways Annotation – CDS Annotation and Refinement video.

The short answer is yes, BLAST is essential. Many computational biologists, geneticists, and numerous other scientists use BLAST day-to-day for complex calculations and analysis. Without BLAST, we would not be able to accomplish most of what is done during Bioinformatic analysis. Could you imagine having to calculate these E-values and Alignments by hand? It isn’t fun. Remember that BLAST is a tool, and like all tools across industries, it is better to stay ahead of the curve and be confident in your ability to use this widely demanded tool.

An e-value, also known as an expected value, is a value provided by BLAST to denote expected random chance when searching the database with your query sequence. The closer your e-value to zero, the less noise we expect to see and the more “significant” our match becomes. Just because we get an e-value of 0.0, doesn’t mean it is perfect though. Let’s examine two situations where an e-value would be 0.0 for different reasons:

• From BLAST, an e-value of 0.0 can occur when we examine short sequences against other short databases that are near identical. According to NCBI, “shorter sequences have a higher probability of occurring in the database purely by chance.” As scientists and annotators, we do not look for chance, we look for evidence to form solid hypotheses and ideas.
• From a computer science perspective, systems hold values as integers and floating point numbers (has a decimal point). When dividing or calculating chances and limits, these floating points, much like on a calculator, have a set amount of space. Because all computers contain what is known as a “Floating-Point Error,” BLAST mitigates this error by rounding to an e-value of 0.0 when our value is < 5*10e-324. To put that into perspective, 5*10e-324 looks like this:

0.00000000000000000000000000000000000000000000000000000000000000000000000 0000000000000000000000000000000000000000000000000000000000000000000000000 0000000000000000000000000000000000000000000000000000000000000000000000000 0000000000000000000000000000000000000000000000000000000000000000000000000 00000000000000000000000000000000005.

• • Simply put, it is because our design of computers in bits is in a Log2 format, this kind of floating-point error is here to stay for a while.
• Overall, e-values that approach 0.0 or 0e0 are far lower than e-values approaching 1. Always think in terms of “noise.” The less noise, the better and more confident we can be in our results.

BLAST results within the GEP should not take several hours. Based on our query sizes, you should expect search times of between seconds to several minutes only. If you are getting search times of several hours, we have a few suggestions for this:

• Review what query sequence you have submitted and against what database.
  • If the incorrect database or query is submitted, BLAST can take much longer to fulfill your request. Take a moment to review your query and databases and run them again.
• Peak Times occur within BLAST. It is best to run a BLAST search on a different internet browser tab at specific peak times. This is due to BLAST being a cloud computing service used by thousands of scientists daily. Being one of those scientists, you will experience the very same delay.
  • These peak times vary, but BLAST will indicate when they occur.

For more information on troubleshooting your BLAST results and why these results may be taking so long, feel free to reach out to any of the GEP Virtual TAs or see NCBI’s troubleshooting page.