Mapping computational requirements
The memory requirements of Map Reads to Reference depends on four factors: the size of the reference, the length of the reads, the read error rate and the number of CPU cores available. The limiting factor is often the size of the reference, while the contribution of the other three factors to the total memory consumption is usually small (see below).
A good estimate for the memory required by the base space read mapper to represent a reference is one MB for each Mbp in the reference. For example the human reference genome requires
of memory.
An additional 4GB of memory should be reserved for the CLC Genomics Workbench, and thus the recommended minimum amount of memory for mapping short high quality reads (e.g. Illumina reads) to the human genome is 8GB. However, when mapping long reads with a high error rate, such as PacBio reads, each CPU core can add several hundred MB to the total memory consumption. Consequently, mapping long reads with high error rate on a machine with many CPU cores, can cause a large increase in the memory requirements for all CLC read mappers.