Library prep is where most runs are won

Library prep is where most runs are won, and where most of them are lost, because the instrument sequences the molecules you hand it. A flow cell does not repair a dimer, invent inserts that failed to ligate, or ignore a contaminant that amplified more eagerly than the sample. This overview is the decision path from input nucleic acid to a library you can defend. The wider frame, from that library to a read file, is next-generation sequencing from library to reads.

The choice of what the library represents, a genome, an exome, an amplicon, or a community, comes before the first bead. Read length and chemistry come with it, as how short reads and long reads differ sets out. What follows assumes you already know which of those questions you are buying molecules for.

Fragment, finish the ends, ligate, then choose a size

A short-read genomic library begins as long DNA and must become inserts the flow cell can cluster. Fragmentation is either mechanical, typically sonication in a focused instrument, or enzymatic, a nuclease treatment whose time and temperature the manufacturer sets. Mechanical shearing is relatively even across sequence. Enzymatic fragmentation can prefer some sequences and still be the right class when you lack a sonicator or the kit is built around it. Write down which one you used. They are not silent substitutes.

Broken ends are not ready for adapters. End repair fills and blunts them. A-tailing adds a single adenosine so a T-tailed adapter can ligate in a known orientation. Ligase then joins adapter to insert. Molecules with an adapter on both ends are the only ones a later PCR, if you run one, will amplify efficiently, and the only ones the flow cell is meant to hold. Molecules with one adapter, or none, are lost at that step or become a background you did not plan.

Size selection removes the extremes. Adapter dimers, two adapters ligated to each other, are short and cluster extremely well. If they remain, they take over the run. Fragments that are too long cluster poorly or produce pairs that do not match the analysis you promised. Bead ratios, gels, or a dedicated size-selection cartridge are the method classes. The ratio that keeps a 350-base insert is not the ratio that keeps a 150-base insert. Follow the kit that matches the sequencer, and confirm on a size trace rather than by faith in the pipette.

PCR is optional when input is high enough for a PCR-free library, and it is required when you still need to add full-length adapters or indexes, or when the mass is otherwise too low to load. Every cycle can duplicate a molecule. Duplicates inflate coverage without adding an independent observation of the genome. The argument is developed in coverage depth is not the same as accuracy. Use the cycle count the kit ties to your input mass. A larger cycle number copied from a different input table is how a low-complexity library is manufactured on purpose.

Tagmentation collapses fragmentation and adapter addition into a transposition. It is a legitimate class with its own input range and its own cleanup. Do not insert an end-repair step into it because a shear-and-ligate diagram showed one.

Input QC is where the loss actually happens

Measure the nucleic acid before you commit a kit from the genomics and sequencing catalogue. The wider study design sits on the genomics research pathway. Measure mass with a dye that sees DNA, or RNA, rather than by a UV absorbance alone, because absorbance will count contaminants that are not your polymer. Purity ratios still help: a dirty extract inhibits ligation and PCR even when the dye says the mass is there. Integrity matters as much as mass. A genomic prep that has already sheared to a smear below your target insert cannot be assembled into that insert by gentler pipetting. An RNA library that started from degraded RNA will sequence the fragments degradation created. Extraction choices that determine that integrity are compared in how DNA extraction methods differ.

Quantify again after the prep, and look at the size distribution. Molarity, not nanograms alone, is what loading cares about, because the flow cell binds ends. A concentrated dimer and a dilute proper insert can share a mass and present very different numbers of ends. The pre-run list in what to check before a sequencing run is the place those numbers have to pass or stop the instrument.

The branch you take when yield is low

Low yield is a symptom with several causes. Do not answer all of them with four more PCR cycles.

If the size trace is empty, the input may have been absent, inhibited, or lost on beads. Check the extraction blank and the sample in parallel. A blank that grows a library is contamination or reagent background, and more cycles will make it louder. A sample that is empty while a matched positive control built a library points at that tube: inhibitors, a failed extraction, or a labelling error.

If the trace is a tall dimer and a small insert peak, the ligation saw too much adapter for the inserts present, or the cleanup failed to remove free adapter. Another PCR will prefer the dimer. Clean or size-select, then quantify. If the insert peak is still a minor species, repeat the prep with the adapter:insert balance the kit specifies for this input, rather than sequencing a library you already know is mostly adapter.

If the insert peak sits at the right length and is merely faint, ask whether you have more input DNA of the same integrity. A new prep from more starting molecules raises complexity. Extra PCR on the faint library raises copy number. Those are different outcomes, and only one of them improves the genome coverage you will be able to trust. When you cannot obtain more input, extra cycles may be the only way to load, and the analysis must then mark duplicates and live with the complexity you have. Say that in the record. Do not report the raw depth as if the cycles had been the kit's minimum.

If the fragments are far longer than the plan, fragmentation did not finish. PCR will not shorten them evenly. Go back to the shear or the enzymatic time. If they are far shorter, you over-sheared or the input was already broken. Size selection cannot invent the missing middle of those molecules.

Quiet failures that still produce a FASTQ file

A library can pass a mass check and still be the wrong DNA. A swapped tube at extraction sequences beautifully and answers another sample's question. Indexes, discussed in adapters indexes and barcode hopping, are the only way a multiplexed run keeps those tubes apart, and they do not correct a swap that happened before the index was added.

Over-amplification also looks like success. The size trace is handsome, the molarity is easy to load, and the duplication rate after the run is severe. Unique coverage collapses. PCR errors that occurred in an early cycle are then present on many reads and can look like a variant. That is a prep artefact wearing a high count.

Bead cleanups lose material when the ratio is wrong or the beads dry out. A low yield after an otherwise normal ligation is often a cleanup loss, not a mysterious genome. Elution volume and the kit's drying time are operational details. They belong in the notebook because they explain the branch you took.

GC-extreme genomes and amplicons with odd lengths will not match a textbook trace. Compare the trace to what this DNA can physically be, not only to the figure in a kit guide written for a different organism.

Biosafety stays an institutional call

Library enzymes do not define the risk group of the organism you extracted. A sample may be infectious before lysis, and some lysis conditions do not inactivate every agent. The Laboratory biosafety manual is a reference document. The decision for your room is local. This overview does not authorise clinical testing, and a research library is not a diagnostic worksheet. Keep the extraction blank when the input is small, because reagent contaminants are a known source of reads that have no business in the sample's story.

Heat, volume, and a library changing hands

Enzymatic fragmentation and ligation are low-volume reactions. In a hot laboratory, or in a room dried hard by air conditioning, an unsealed plate changes concentration while you prepare the next row. Seal it. Humidity in a monsoon week puts condensate on a cold reagent brought straight to the bench; let it equilibrate and wipe the outside before you open it, so the water on the tube is not part of the reaction. A finished library that must travel between buildings needs a cold handoff written into the specification, not an afternoon in traffic at ambient heat. Record the temperature the tube was held at. A later duplication rate will not tell you that the library sat warm.

What the enquiry should carry

State the organism, the input type and integrity, the library class (shear and ligate, tagmentation, PCR-free or amplified, amplicon or capture), the insert length you need, and the index design. Say what you already measured. A request that names only the sequencer has skipped the step that decides the run.

Whole-genome library choices can be discussed against the whole-genome sequencing enquiry reference. A prep that exists only to check one amplicon can be discussed against the Sanger DNA sequencing enquiry reference, which may remove the need for a library at all. Send the input measurements with the quote request. Writing the acceptance checks so a low yield cannot be redefined as success after the fact is part of commissioning a sequencing or proteomics study.