# BLAT and in-silico PCR **TL;DR:** The `blat` plugin adds two **Tools** menu items that answer "where is this sequence in the genome", **BLAT search** and **In-silico PCR**. Both query UCSC-style sequence-search servers and turn the hits into a track built on the fly, with every hit listed so you can go to the one you want. The plugin ships with JBrowse Desktop, where both items are in the Tools menu. JBrowse Web does not bundle it yet, so the menu items are absent there; shipping it on Web is planned. ## BLAT search Choose **Tools → BLAT search…**, paste a DNA sequence or FASTA, pick the assembly, and submit. What comes back is a new track named after the query, holding the hits, and a **Search results** panel listing them, ranked by UCSC's `pslScore`. Each location is a link that takes the view there, labelled with the percent identity and the fraction of the query the hit covers. A query adds the track and shows you the list, leaving the view where it was. Which hit matters is yours to decide, and the top-scoring one is not always it. The track is an alignments track. A PSL hit is a list of aligned blocks in query and target coordinates, which is what a CIGAR encodes, so each hit is drawn the way a read is: blocks as aligned runs, a target gap as a deletion, a query gap as an insertion, and the unaligned ends of the query as soft clips. Because the sequence you submitted is known, the hit also carries it, and the pileup marks every base that disagrees with the reference. That is the difference between a hit that is your sequence and one that merely scores well. BLAT returns every placement of a query. The best-scoring one is primary and the rest are marked secondary, so competing mappings of one sequence stay off the best hit's row.
Pasting FASTA searches each record separately, up to 25 records and 25 kb of sequence in total. Records keep their own names, so hits from different queries stay attributable in the track. A bare sequence with no header is named `YourSeq` by the server. ### Which database gets searched The assembly picker drives the UCSC database, resolved in this order: - the db stamped on the assembly's `sequence.metadata.blatDb`, when there is one - a small alias map for the common names that are not literal UCSC db ids (`GRCh38` to `hg38`, `T2T-CHM13v2.0` to `hs1`, and similar) - the assembly name as typed **Show advanced settings** exposes the resolved db so you can override it, including with a GenArk accession such as `GCF_000001405.40`. An assembly UCSC does not host has nowhere to search — a genome you opened from your own files is the usual case — and the dialog says so before you submit. If UCSC does have a database for it, set that database under advanced settings. These dialogs have no local aligner behind them; [](https://jbrowse.org/jb2-staging/docs/user_guides/sequence_search) scans the reference for an exact pattern without involving UCSC. ### CAPTCHA and apiKeys The UCSC BLAT server is length-limited, and public requests to it are gated behind a CAPTCHA. Both dialogs default to a jbrowse.org proxy that supplies an apiKey server-side, so an ordinary search meets neither the CAPTCHA nor a key — at the cost of a request budget every user of that proxy shares. You meet the CAPTCHA by pointing the server field at UCSC yourself, or on desktop when the shared budget is spent and the search falls back to UCSC. Solve it in the window that opens, or paste a UCSC apiKey (generate one from a UCSC Genome Browser account under Hub Development → API key) to skip it. On desktop, entering a key moves the server field to UCSC, because the proxy replaces a key you send with its own — so a key only spends against UCSC directly. ## In-silico PCR Choose **Tools → In-silico PCR…** to find where a primer pair amplifies. Enter a forward and reverse primer and an optional maximum product size. This uses UCSC's `hgPcr` service and follows the same database-selection and apiKey/proxy options as BLAT search. Predicted amplicons arrive as a track, with the same **Search results** panel listing every product with its size and primer pair. As with BLAT, clicking a product is what takes the view to it. A PCR product has the shape of a paired-end read: two short oriented footprints pointing at each other with an unsequenced insert between them, so each product is drawn as a read pair with view-as-pairs on. The two arrows converge whatever strand the product is reported on, because a primer's direction is its own and not the amplicon's, and the line between them is the interior you never sequence. Because the primers themselves are carried as the reads' bases, a base where a primer disagrees with the template is drawn as a mismatch. UCSC tolerates one toward a primer's 5' end and still reports the product, which is what a primer sitting over a SNP looks like; it requires the last 15 bases at the 3' end to match exactly, and reports no product otherwise. More than one product means more than one band. The results panel counts the products and lists each size, which is what you compare a gel against, and says so explicitly when there is more than one. Two products of similar size will not resolve.
## Notes Result tracks store their features inline in the session, so they live as long as the session does and travel with it when you save. Each search adds another track; delete the ones you are done with from the track selector. ## See also - [](https://jbrowse.org/jb2-staging/docs/user_guides/sequence_search) - [](https://jbrowse.org/jb2-staging/docs/user_guides/sequence_track)