Project Computation Sub-Tab

2.7.2

The Computation sub-tab is shown on Figure 57. It allows you to adjust the settings described below.

Figure 57: gsMapper Parameters tab, Computation sub-tab for Genomic projects

•

Incremental reference mapper analysis – Results from computations performed with this option selected (checked) will serve as the basis for the mapping of additional reads. If a project computation is performed with this option deselected, all the project data will be mapped anew each time it is computed, i.e. forcing all alignments to be recalculated and overwriting any existing results.

◦	Default: selected (Note: This setting is not saved with a project, and is always reset to “selected” when a project is re-opened).

•

The number of CPUs option can be given to limit the load of the software on the computing resources. The default is 0, which specifies that all of the CPUs on the computer should be used. The computation uses the CPU setting to create an approximate load on the computer, so the actual load may vary somewhat from this number, i.e., using “-cpu 3” may cause a load from 2 to 4 on the system.

Currently, only the “Computing alignments…”, “Mapping short reads to reference …”, and “Generating output…” phases have been parallelized (because they are the most computationally intensive phases for larger projects). Other phases of the computation will still use single threading. Also, note that only CPUs which have access to the same physical memory can be used.

The current memory footprint is roughly 4 bytes per reference base plus 2 bytes per input (read) base. For example a 1 Gbp genome with a 20x coverage would need 44 GB of physical memory.

•

Use duplicate reads – when checked, the Mapper will include duplicate reads when computing the consensus for a contig and when computing frequency statistics for 454AllDiffs.txt and 454HCDifffs.txt. Duplicate reads are a known potential artifact of the emPCR Amplification process which may occur, for example, when two or more beads are amplified in a single emulsion microreactor. In such cases, duplicate reads should be excluded from the computation, which is the default. This option should be used in certain contexts, e.g. when analyzing an Amplicon sequencing experiment using the GS Reference Mapper, where every read is a separate datapoint.

•

Use serial I/O – select this option if you are using many .sff files (> 4 million reads) in your project. The option instructs the software to prepare an intermediate read file in which the reads appear in the same order as they do in the alignments. For large projects, this can speed up the execution of the Mapper, especially during the Output Phase of its computation. For more information, see Section 4.6.

•	Overlap Detection Parameters:

Parameter	Description	Default Value	Allowed Values
Seed step	The number of bases between seed generation locations used in the exact k-mer matching part of the overlap detection.	12	>= 1
Seed length	The number of bases used for each seed in the exact k-mer matching part of the overlap detection (i.e. the “k” value of the k-mer matching).	16	6 - 16
Seed count	The number of seeds required in a window before an extension is made.	1	>= 1
Hit-per-seed limit	A filtering limit on the seeds found during the exact k-mer matching (if more than this number of seed matches are found for a lookup position in the query sequence, those seed matches are not tested as candidate alignments). If more than 70% of a read’s seeds have more than this number of seed hits, the read is not used and is classified as Repeat.	70	>= 0
Minimum overlap length	The minimum length of overlaps used by the mapper. This value can be entered as either a number of bases or a percent of read length. To enter a percent, the % symbol must be used (e.g. 37%).	40	>= 1 (length) 1 – 100 (percent)
Minimum overlap identity	The minimum percent identity of overlaps used by the mapper.	90	0 - 100
Alignment identity score	When overlaps are extended during an alignment, this is the per-overlap-column identity score used to calculate the overall score for the alignment.	2	>= 0
Alignment difference score	When overlaps are extended during an alignment, this is the per-overlap-column difference score used to calculate the overall score for the alignment.	-3	<= 0
Repeat score threshold	The threshold used to declare alignments “unique” or “multiple”. If an input read aligns to more than one position in the reference sequence(s), an alignment score is calculated for each alignment by summing the identity or difference score for each alignment column; if the best alignment score is greater than the score for any other alignment by more than this threshold value, it is marked as “uniquely mapped”. Otherwise, it is “multiply mapped” or “Repeat”.	12	>= 0

When a read is mapped, exact k-mer matches between the read and the reference yield a set of overlaps between the two. These overlaps are then extended into longer alignments. The best alignment is used to determine the mapping location of the read. In some cases, non-overlapping portions of a read may map to different loci in the reference(s). Together these mappings may form a chimeric best alignment for the read. Chimeric mappings favor alignments that are more closely placed along a single reference.