Efficient and Practical Sample Pooling High-Throughput PCR Diagnosis of COVID-19

Haran Shani-Narkiss; Omri David Gilday; Nadav Yayon; Itamar Daniel Landau

Here we present two pooling strategies that offer to dramatically reduce the use of such resources, as well as time and labor. In regions and testing conditions in which positive tests are very rare (p<0.05), a strategy of repeated pooling can be extremely efficient by first selecting an initial batch size that yields probability 0.5 of being entirely negative, and then proceeding by positive batches in half at each stage. As mentioned above, when positive samples are exceedingly rare this strategy in principle calls for very large batch sizes, well into the hundreds and even thousands. Such large batches are unfeasible with existing protocols. However, as we have shown, the strategy of repeated pooling is highly efficient in settings of exceedingly rare positives even when batch size is constrained to a pragmatic limit such as 64. Nevertheless, the process of repeatedly splitting pools into two may be challenging for many laboratories to implement in practice, and it loses marginal efficiency as the frequency of positive tests increases. We therefore show that a simpler protocol of one-time pooling, with optimized initial batch sizes is very efficient for all p up to about 0.2. One-time pooling is efficient even when the size of possible initial batch sizes is technically limited, for example to 16, either in order to simplify laboratory protocol or because knowledge of p is lacking. We show that both methods compare favorably to fixed-size matrix methods, that attempt to exploit 2D positional information from samples arranged on a grid. We note, however, that matrix methods can in principle also be optimized for given p, or by using complex pooling strategies 10 . Optimized matrix methods may prove more efficient than the two straight-forward methods presented here.