How to SPARQL — a step-by-step tutorial

Before we start: the data is made of little sentences

SABIO's data is stored as millions of tiny three-part sentences: something → has some property → some value. For example:

organisms/5          is a                 Organism
organisms/5          has the name         "Homo sapiens"
kinlaw-entry/23815   was measured in      organisms/5

The first part is always a thing, written as a web address (organism 5, entry 23815). The value at the end is either plain text or a number, or another thing — that's how entries are connected to their organism, enzyme, publication and so on.

A query is a sentence with blanks. You write the sentence, leave a blank where you don't know the answer, and SPARQL fills in every possible answer. A blank is a word starting with ?, like ?name.

Step 1 · Peek at five sentences

The simplest query there is: a sentence that is only blanks.

What's going on

• WHERE { … } holds the sentence you're looking for.
• ?thing ?property ?value — three blanks, so every sentence matches.
• SELECT * means "show me all the blanks".
• LIMIT 5 means "five answers are enough". There are millions of sentences, so always keep a LIMIT while you explore.

You'll probably see sentences about unit 0 ("no unit"). It just happens to come first; without more detail in the sentence, SPARQL returns whatever it finds first.

Step 2 · Everything about one thing

Now fill in the first part and keep the other two blank: "tell me everything about organism 5".

What's going on

• A thing's full address goes in < >. It's also the address of its page on the SABIO-RK website.
• The answer lists organism 5's properties: what it is (type → Organism), its name (label), other names (altLabel), the same organism in the NCBI taxonomy (exactMatch) and how many entries use it.
• Here we SELECT just the two blanks we care about, instead of *.

Step 3 · List all organisms

Two sentences at once: "?organism is an organism" and "?organism has the name ?name".

What's going on

• PREFIX lines are abbreviations: sabio:Organism is short for <https://sparql.sabiork.h-its.org/vocab#Organism>. Just copy them to the top of every query.
• a is short for "is a".
• rdfs:label is the name. Every SABIO thing has one.
• Each sentence ends with a dot .
• Both sentences use the same blank ?organism, so they must be about the same organism. That's how you connect sentences.

Step 4 · Find something by its name

Usually you know the name, not the address. Put the name in quotes and leave the thing blank:

What's going on

• Text in quotes must match exactly, including upper and lower case. "homo sapiens" finds nothing.
• The answer is the address organisms/5 we used in step 2.

Step 5 · Search for part of a name

Don't know the exact name? Keep the name blank and add a FILTER: a condition every answer has to pass.

What's going on

• LCASE(?name) turns the name into lower case, so the search ignores upper/lower case.
• CONTAINS(…, "coli") keeps only names that contain "coli".

Step 6 · Sort the answers

Which organisms have the most entries? Every organism carries its entry count, so we sort by it.

What's going on

• A semicolon ; means "same thing again". These three lines are the same as writing ?organism at the start of each one. It just saves typing.
• ORDER BY DESC(?entries) sorts from biggest to smallest. Without DESC it sorts smallest first.

Step 7 · Follow a link from one thing to another

This is where it gets useful. An entry points to its organism and its enzyme, and the enzyme has a name. We follow that chain: Homo sapiens → its entries → their enzyme → the enzyme's name.

What's going on

• Read it top to bottom, like a path: find the organism, then entries measured in it, then each entry's enzyme, then that enzyme's name.
• Each blank links two sentences: ?organism links lines 1 and 2, ?entry lines 2 and 3, ?enzyme lines 3 and 4.
• Every ?entry is the address of an entry's page.

Step 8 · Count

Same chain as step 7, but instead of listing entries we count them per enzyme: "which enzymes have the most human entries?"

What's going on

• GROUP BY ?enzymeName puts all answers with the same enzyme name in one pile.
• (COUNT(?entry) AS ?entries) counts each pile and calls the number ?entries.

Step 9 · Get the measured values

The last step: the actual numbers. An entry has parameters (Km, kcat, Ki, …), and each parameter has a type, a value, a unit and usually a compound. Here: Km values of alcohol dehydrogenase (EC 1.1.1.1) in Homo sapiens, per substrate.

What's going on

• Nothing new — just more sentences chained together. The empty lines are only there to make it easier to read.
• Values are converted to standard units, which is why Km shows up in M (molar).
• sabio:ecNumber "1.1.1.1" finds the enzyme by its EC number instead of its name.

That's it

With these nine steps you can already answer most questions: describe the sentence you're looking for, leave blanks for what you want to know, and connect things through shared blanks. A few things to remember:

• Keep a LIMIT while exploring.
• Names in quotes must match exactly; use FILTER(CONTAINS(LCASE(…), "…")) when you're not sure.
• If a query returns nothing, remove sentences from the bottom until it does — then you've found the one that doesn't match.

Ready for more? The query editor has a list of explained example queries on the left, from simple to advanced.