Using MPI with C\# and the common language infrastructure

Abstract

We describe two interfaces for using the Message Passing Interface (MPI) with the C# programming language and the Common Language Infrastructure (CLI). The first interface provides CLI bindings that closely match the original MPI library specification. The second library presents a fully object-oriented interface to MPI and exploits modern language features of C#. The interfaces described here use the P/Invoke feature of the CLI to dispatch to a native implementation of MPI (in our case, LAM/MPI). Performance results using the Shared Source CLI demonstrate there is only a small performance overhead incurred.

Integration of Clustering and Multidimensional Scaling to Determine Phylogenetic Trees as Spherical Phylograms Visualized in 3 Dimensions

“…2,2013), and none of the tests had p-values grea suggesting all of the measured correlation significant despite the increased type I error ( rate that can occur with Mantel tests [43]. The very high correlations betw matrix and the MSA distance ma visualization regardless of sequ because the input alignment is i and only the distance calculation pairwise alignments for the same correlations with the RAxML r they still provided a reasonably go The relationships between gen phylogenetic tree created with consistent with the current und phylogenetic relationships [11] Racocetra, Scutellospora, and Gi to the same evolutionary gro phylogenetic tree ( Figure 5) and possible to visualize the how correlate with the sequences after required in the interpolated tree that are the same color in the M tree does not match well with because the sequences on th phylogenetic tree are more sim other sequences in the dataset, an ve comparable correlation orter sequences (dataset 2). ween the RAxML reference atrix used for MDS cluster uence length are expected identical for both matrices n method is different.…”

“…Finally, we added the repre this trimmed alignment using MAF create dataset 2. This gave a MSA f 801 sequences from [11] and 505 for a total of 1306 sequences, and 454 optimized) with: 514 sequence from GenBank, and 126 represent total of 1020 sequences. For this ph we selected a smaller set of sequen expected range of genetic variabil RAxML took about 4 hours to finis hours to finish on the second data MDS only took a few minutes to using same amount of cores.…”

“…We applied 6 clusters and then picked a uster as part of the dataset. enBank were added to the [11] using MAFFT [42]. In quence lengths affected the etic trees and clustering, we uences that shared the same gene: one dataset contained ontained shorter sequences.…”

“…only retained the unique ded region beyond the D2 es long without gaps); then the unique sequences that site and the average end ghly 425 bases long without sentative 454 sequences to FFT as described above to for dataset 1 (999nts) with: sequences from GenBank for dataset 2 (599nts with s from [11], 380 sequences tative 454 sequences for a hylogenetic comparison test nces that still represents the lity within AM fungi. The sh on the first dataset and 7 set using 8 cores.…”

“…All sequences were from species of arbuscular mycorrhizal (AM) fungi because they exhibit a large amount of sequence variation both between species as well as within species [10], which can make them challenging to analyze [11]. The sequence datasets were derived from a combination of: (1) a large-scale AM fungal phylogenetic study [11]; (2) additional sequences obtained from GenBank to increase the taxonomic coverage of the dataset; (3) representative 454 pyrosequences from spores of known AM fungal species that were selected using DACIDR [8]. DACIDR uses pairwise clustering and MDS for robust and scalable sequence clustering and visualization for more than one million sequences [12].…”