Is it possible to pool different library types in the same sequencing run?

Different library workflows can produce libraries that differ in terms of size, index length, clustering efficiency and more. Illumina recommends pooling together only libraries prepared with the same library preparation workflow for sequencing. This recommendation applies to libraries pooled in the same run (for flow cells without individual lane loading) or in the same lane (for flow cells that use individual lane loading) for the following reasons.

1. Unpredictable cluster density and negative impact on sequencing quality. Different library types can have different sizes and shorter libraries cluster more efficiently than longer libraries. Therefore, smaller libraries will be overrepresented compared to larger libraries (see Figure 1), potentially leading to an overclustered run. If the number of cycles during a sequencing run exceeds the length of the shorter library, run quality is negatively impacted.

Figure 1: An equal ratio of two different library lengths in a pool can lead to overrepresentation of the shorter library. 2. Uneven clustering and data output. Different library types can have different clustering efficiencies, even when they are similar in size. This difference can lead to unpredictable clustering/cluster density and difficulty determining the best loading concentration to use. Illumina recommends running a single library type and optimizing the loading concentration for the specific sequencing instrument based on the following guidance: Clustering Optimization Overview Guide How to achieve more consistent cluster density on Illumina sequencing platforms Furthermore, different library types have different read depth/coverage requirements. For example, a whole-genome sequencing library can require more coverage than an amplicon library (see Figure 2). The loading concentration for different library types must be optimized to achieve the correct level of coverage, and Illumina does not provide guidance for this optimization. The coverage calculator tool can be used to establish the number of samples per run for a single given library type.

Figure 2: An equal ratio of two different library types in a pool can lead to overrepresentation of one type even if they are the same length, which can affect the read distribution and coverage. 3. Difficult and possibly ambiguous demultiplexing. Different library types use different types of indexes (eg, single- versus dual-indexed and different index lengths). Set up a run (or lane for flow cells with individual lane loading) to read either single or dual index reads, but not both. Set Index Reads to a single length for both i5 or i7 index (Figure 3).

Figure 3: The red X’s indicate index positions where an index cannot be read for all libraries in the pool. Green checks indicate positions where the index can be read for all libraries. In this pool, the indexes are reliably read as only a single, 6 bp Index Read because the i7 index is read first and is the only Index Read for single index libraries. Such a run setup can negatively impact demultiplexing.

In conclusion, pooling different library types on the same run or lane when not supported requires independent optimization and validation.

NOTE: It is possible to pool different library types on flow cells supporting individual lane loading if the different library types are loaded on separate lanes. The NovaSeq Xp Workflow, the HiSeq Rapid Duo cBot Sample Loading Kit, and the HiSeq 2500 High Output modes all support individual lane loading. However, only one read length configuration can be used for a single sequencing run, even with separate flow cell lanes.

For any feedback or questions regarding this article (Illumina Knowledge Article #3284), contact Illumina Technical Support [e-mail].