How do Researchers Search for Scientific Resources: the Traditional (but still useful) Method? (1/3)

A researcher rarely begins by asking, “Which website should I visit?”
The starting point is usually a scientific question, a practical task, or finding a protocol.

For example:
“I need to identify proteins expressed in activated microglia.”
Or:
“I need software for detecting off-target effects in CRISPR experiments.”
Or:
“I need an antibody validated for detecting POU5F1 in human cells.”

It is easy to realize that these questions contain several concepts at once: a biological entity, an experimental context, a method, an organism, and sometimes a desired result.

The researcher therefore translates the scientific problem into searchable concepts.
A search for “microglia RNA-seq,” for example, may evolve into:
“single-cell RNA-seq microglia differential expression”
and then into:
“single-cell RNA-seq differential expression in immune cells.”

This reasoning is normal. Searching for information is part of the research process itself: researchers learn new terminology from papers and tools while searching, and then use that terminology in queries.

Scientific language is precise, but unfortunately, it is not always uniform.
A single biological entity can have an official name, abbreviations, historical names, synonyms and database-specific identifiers. UniProt, for example, maintains protein names, alternative names and gene names because a protein may be described in several ways. It also provides advanced search facilities for restricting queries to particular fields and combining terms with Boolean logic.

Let's take the example of POU5F1. A searcher can find:

• POU5F1
• OCT4
• Oct-4
• POU domain, class 5, transcription factor 1

These terms can refer to the same biological entity, depending on context.

This same problem arises in many other scientific fields, such as diseases, organisms, compounds, experimental techniques, and software. Therefore, a scientific search strategy does not rely on a single keyword, but uses synonyms and/or identifiers.

PubMed, for example, uses the MeSH controlled vocabulary to help address differences in terminology between authors and researchers. Its search system also supports Boolean operators, field-specific searches and phrase searching.

There is really no single universal search engine for science.
AI tools have recently been developed for scientists; they will be the subject of another article.
A researcher may begin with a general web search engine (for example, Google or Bing) to discover possible resources. The search may then move to a specialized literature website, like PUBMED, a biological database or a scientific organization's website.

For example, a protein-related investigation might involve:

• Google or another general search engine
  Useful for discovering unfamiliar resources or finding a website whose name is already known.
• PubMed
  Useful for finding biomedical literature and publications indexed in MEDLINE and related resources.
• Europe PMC
  Useful for searching biomedical literature and related research outputs, with advanced tools for refining searches and even restricting them to article sections or data availability information.
• UniProt
  Useful for finding proteins by name, identifier, biological characteristics, and sequence, with both free-text and field-specific search capabilities.
• Specialized registries and directories
  Useful when the researcher is looking for a tool, database, service or website rather than a publication.

This multi-channel "search approach" is logical for scientific resources. A general search engine can help a researcher discover a concept, while a specialized resource will be where the information is located.

The search for information is often iterative.
Imagine a researcher needs software to analyze single-cell transcriptomics data. The first query might be:

• “single cell RNA software”
  The results may be too broad. The researcher may then add the experimental objective:
• “single cell RNA differential expression software”
  The search might become even more specific:
• “single cell RNA differential expression immune cells”

Eventually, the researcher may search for the name of a tool, a method, a dataset, or a paper describing an algorithm.
This is closely related to how to conduct a search for information.
PubMed allows researchers to combine terms with AND, OR and NOT, search specific fields and use MeSH terminology. Europe PMC likewise provides advanced query capabilities and allows users to refine searches by content type, metadata and article sections.

The important lesson is that the first query is often only the beginning.
For website owners, this means that a page should not be optimized around a single trendy keyword. It should clearly communicate the broader scientific concepts related to the resource.