A BLAST search of a new sequence returns top hits to cloning vectors. What would you do?

Why

NCBI defines a contaminated sequence as one that does not faithfully represent the genetic information of its source because it contains segments from elsewhere. Vector, adapter, linker and primer sequences are common sources, because constructs are cloned, propagated and amplified before sequencing. Leaving them in produces misleading search results and, if submitted, contaminated database entries.

VecScreen searches a query against UniVec, a database built to hold one copy of every unique segment from many vectors plus common adapters, linkers and primers. It grades matches by how often a random match would occur, with terminal matches treated differently from internal ones. A strong match is one expected about once in a million queries of 350 kb.

Where the vector sits matters. Matches at the ends usually mean incomplete trimming of cloning or sequencing adapters. An internal match can mean a chimeric clone or an assembly that joined vector to insert.

UniVec has limits its own documentation states: because shared segments are stored once, it will not identify which vector matched best. The cloning history of the sample is the way to pin that down.

How to work it through

1. Run VecScreen on the sequence and note each match's strength and position.
2. Compare the matching segments with the cloning vector and adapters recorded for the sample.
3. Trim terminal vector and adapter segments.
4. Investigate internal matches as possible chimeras or misassembly.
5. Search again with the trimmed insert only.
6. Screen the rest of the batch to see whether the problem is systematic.

Where the answer stops holding

VecScreen only finds sequences that are in UniVec, so new or custom constructs can be missed. Some genuine genomic sequences resemble vector parts, such as viral or bacterial elements. Contamination from other organisms needs a separate screen, such as NCBI's foreign contamination screen.