How should adapter dimers be optimized when they are too high after微量cfDNA library preparation?

After cfDNA library preparation with 50 pg input, if you see a peak around 150 bp, do not immediately conclude it is adapter dimer. cfDNA fragments themselves are concentrated at 160-170 bp, and with adapters added, they may reach around 250-280 bp. However, if the original fragments before library preparation are on the shorter side, the library peak will be at 150-180 bp. Adapter dimers are usually closer to 120-150 bp and have a sharp peak shape. It is recommended to use an Agilent 2100 or capillary electrophoresis to check whether the peak is consistent with a no-adapter control, and to perform qPCR or validate with Y-shaped adapter-specific primers. If it is confirmed to be dimers, then address adapter input amount, ligation efficiency, and purification ratio, rather than simply changing the kit.

From our own comparison, those with better adapter dimer control include KAPA HyperPrep, NEBNext Ultra II, as well as domestic micro cfDNA kits from Vazyme and Diying, but consistent performance across all batch samples cannot be guaranteed. It is recommended to take the same 50 pg cfDNA, aliquot it, and simultaneously run small-scale tests with 2-3 vendors, focusing on library yield, dimer ratio, insert size, and Q30. Before purchasing, request trial packs or small packaging, which many distributors can provide. Also note that different kits have different designs; some use uracil-containing adapters to reduce dimers, while others rely on bead ratios; rather than switching kits, first clarify your own workflow variables.

A high proportion of adapter dimers usually indicates excessive adapter or low ligation efficiency. With such a low input amount as 50 pg, if the adapter is added at the standard amount, free adapters tend to self-ligate and Y-shaped adapters tend to ligate to each other. You can try diluting the adapter to 1/2 or 1/4 of the standard concentration while keeping the ligase and buffer unchanged. Using an imported kit, after adding 1/4 adapter at 50 pg, the dimer proportion decreased significantly, and the main peak was no longer the dimer. Also check the ligation reaction temperature and time; overnight ligation is not necessarily good. Some kits are suitable for 16°C for 20 minutes, and longer times increase dimers. Changing the adapter dilution ratio is one of the fastest variables to verify.

A starting amount of 50 pg is close to the lower limit of many kits. The manual states 10 pg sensitivity, but that does not mean all samples can yield good libraries at the same cycle number. When the input is low, it is easy to increase the cycle number to obtain sufficient concentration, and as a result, adapter dimers are amplified simultaneously. It is recommended to first perform qPCR to monitor the amplification curve and stop before the plateau is reached, usually reducing by 1-3 cycles compared with the default. When we work with cfDNA, for 50 pg input we start with 10-12 cycles and stop before the plateau is fully reached. After lowering the cycle number, the dimer proportion improves markedly, and the library concentration can still meet QC requirements.

You have already switched to fresh magnetic beads, but if the ratio is still based on the manual, it may not be optimized for dimers. Generally, after low-input cfDNA library preparation, two rounds of SPRI are performed: in the first round, use 0.9X–1.0X magnetic beads to remove large fragments/unligated products, and retain the supernatant; in the second round, use 0.7X–0.8X magnetic beads to recover the target library, leaving dimers in the supernatant. The specific ratio should be fine-tuned according to your library fragment distribution. It is recommended to first take one library preparation product and divide it into several tubes, test combinations of 0.7X/0.8X/0.9X, and run a gel to assess dimer removal efficiency. Note that the magnetic beads should be fully equilibrated to room temperature before use, and the washing steps should be gentle to avoid loss of the low-input library.