1,721,036 research outputs found
anneflo/Clusters_resistance_spread: v1.0
No full textMatlab implementation of the code described in "Antibody-mediated cross-linking of gut bacteria hinders the spread of antibiotic resistance", by Florence Bansept, Loïc Marrec, Anne-Florence Bitbol and Claude Loverdo
Performance of the MI-IPA without a training set.
No full text(A) Final TP fraction obtained by the MI-IPA and the DCA-IPA versus Nincrement for the standard HK-RR dataset. At the first iteration, the CA is made of random within-species HK-RR pairs. At each subsequent iteration n > 1, the CA includes the (n − 1)Nincrement top predicted pairs. All results are averaged over 50 replicates employing different initial random pairings. The MI-IPA was also run on an alignment where each column is randomly scrambled. The dashed line represents the average TP fraction obtained for random within-species HK-RR pairings. (B) Final TP fraction obtained by the MI-IPA and the DCA-IPA versus the total number M of HK-RR pairs in the dataset, starting from random pairings. For each M, except that corresponding to the full dataset, datasets are constructed by picking species randomly from the full dataset, and results are averaged over multiple different such alignments (from 50 up to 500 for small M). For the full dataset (largest M), averaging is done on 50 different initial random within-species pairings. All results in (B) are obtained in the small-Nincrement limit.</p
Performance of the MI-IPA for different protein pairs.
No full textFinal TP fraction obtained without a training set by the MI-IPA and the DCA-IPA versus Nincrement. (A,B,C) Pairs involved in ABC transporters; datasets of ∼5000 pairs extracted from larger full datasets. (D,E) Pairs involved in enzymatic complexes; full datasets of ∼2000 pairs. In each case, the mean number 〈mp〉 of pairs per species is indicated, and dashed lines represent the average TP fraction obtained for random within-species pairings. All results are averaged over 50 replicates that differ in their initial random within-species pairings.</p
Inferring interaction partners from protein sequences using mutual information
Get PDFFunctional protein-protein interactions are crucial in most cellular processes. They enable multi-protein complexes to assemble and to remain stable, and they allow signal transduction in various pathways. Functional interactions between proteins result in coevolution between the interacting partners, and thus in correlations between their sequences. Pairwise maximum-entropy based models have enabled successful inference of pairs of amino-acid residues that are in contact in the three-dimensional structure of multi-protein complexes, starting from the correlations in the sequence data of known interaction partners. Recently, algorithms inspired by these methods have been developed to identify which proteins are functional interaction partners among the paralogous proteins of two families, starting from sequence data alone. Here, we demonstrate that a slightly higher performance for partner identification can be reached by an approximate maximization of the mutual information between the sequence alignments of the two protein families. Our mutual information-based method also provides signatures of the existence of interactions between protein families. These results stand in contrast with structure prediction of proteins and of multi-protein complexes from sequence data, where pairwise maximum-entropy based global statistical models substantially improve performance compared to mutual information. Our findings entail that the statistical dependences allowing interaction partner prediction from sequence data are not restricted to the residue pairs that are in direct contact at the interface between the partner proteins.</div
anneflo/MI_IPA: First release
No full text<p>This is the first release of the source code of the MI-IPA.</p>
Signature of protein-protein interactions.
No full textDistribution of the fraction of MI-IPA replicates where each possible within-species protein pair is predicted as a pair, for three different pairs of protein families. The MI-IPA replicates only differ in their initial random pairings. (A and B) Interacting protein families: BASS-BASR homologs and MALG-MALK homologs. (C) Protein families with no known interaction (BASR-MALK homologs). In all panels, the distribution of replication fraction obtained on actual sequence alignments is shown together with the same distribution obtained by running the MI-IPA on alignments where each column is randomly scrambled (null model). All alignments include ∼5000 protein pairs, with average number of pairs per species 〈mp〉 ≈ 5, and each distribution is estimated from 500 MI-IPA replicates, using Nincrement = 50.</p
anneflo/Mirrortree_IPA: First release
No full text<p>This is the first release of the source code of the Mirrortree-IPA.</p>
Performance of the MI-IPA for different training set sizes <i>N</i><sub>start</sub>.
No full text(A) Increase of the TP fraction with the number of iterations during the MI-IPA. (B) Initial and final TP fractions (at the first and last iteration) versus Nstart for the MI-IPA and for the DCA-IPA of [21]. In both panels, the standard dataset of HK-RRs is used, and the CA includes Nincrement = 6 additional pairs at each iteration. All results are averaged over 50 replicates that differ by the random choice of HK-RR pairs in the training set. Dashed lines represent the average TP fraction obtained for random within-species HK-RR pairings.</p
Increase of pairwise MI obtained by the MI-IPA.
No full text(A) Estimates of the pairwise MI (sum of MIs of all inter-protein residue pairs) are shown versus the total number M of sequences in the dataset. The different curves correspond to the pairwise MI of the actual set of correctly paired HK-RRs, of the initial random within-species assignment used to initialize the MI-IPA, and of the final assignment predicted by the MI-IPA, as well as of an alignment where each column was scrambled. Both axes have a logarithmic scale. The slope −1, expected from leading-order finite-size effects, is indicated by the dashed line. For each M, HK-RR datasets are constructed by picking species randomly from the full dataset, and results are averaged over 50 different such alignments. (B) Excess pairwise MI of the actual set of correctly paired HK-RRs, compared to the initial random within-species pair assignments and to the final assignment predicted by the MI-IPA. The MI-IPA successfully reduces this excess pairwise MI, thus approaching the pairwise MI of the actual alignment. Same data as in (A).</p
- …
