Return to search

Prediction of Ecological Function in the Microbiome Using Machine Learning on the Graph Spectra of Coevolving Subnetworks

<p> <b>Chapter 1.</b> We propose a method for predicting the ecological function of host-associated microbes using neural networks trained on a feature space of labeled ecological interactions from the literature. The feature space is constructed over the Laplacian spectral density distributions of the networks formed by linking the phylogenies of the host and microbial clades through their ecological associations. A classifier trained on 51 interactions with known ecology and 100 simulated controls was used predict the ecological function in the microbimes of 14 species of wild-caught cichlid fish from the Lake Tanganyika adaptive radiation observed using 16S rRNA sequencing. </p><p> <b>Chapter 2.</b> Genomic patterns of divergence are examined using whole-genome resequencing of three sympatric cichlid species pairs with similar functional and ecological differentiation but different ages, revealing a signature of genomic divergence. Regions of elevated relative differentiation exhibit increased absolute differentiation. We detect a signature of convergent evolution across all three species pairs. Our results suggest that functional phenotypic differentiation is associated with a signature of genomic divergence. </p><p> <b>Chapter 3.</b> We show that evolutionary innovations can result in competitive inferiority and extinction. The modified pharyngeal jaws of cichlid fishes and several marine fish, a classic example of evolutionary innovation, are not universally beneficial. Analysis of dietary evolution across marine fish lineages reveals that the innovation compromises access to predator niches. This competitive inferiority shaped the adaptive radiation of cichlids in Lake Tanganyika and played a pivotal, previously unrecognized role in the mass extinction of cichlid fishes in Lake Victoria after Nile perch invasion. </p><p> <b>Chapter 4.</b> We map protein sequences from the Global Ocean Survey to protein families and use non-negative matrix factorization (NMF) to approximate linear combinations of ecological components with characteristic functional and site profiles. We identify functional signatures, estimate functional distance between sites, and find that an NMF-filtered measure is more strongly correlated with environmental distance than a comparable PCA-filtered measure. We find that functional distance is more strongly correlated with environmental distance than geographic distance in agreement with prior studies. </p><p> <b>Chapter 5.</b> We describe the unique technical, logistical, organizational, and ethical issues from the 2013 Indigo V Indian Ocean Expedition research cruise from Cape Town, South Africa, to Phuket, Thailand aboard the S/Y <i>Indigo V.</i> An inventory the surface water population of bacterioplankton was collected and basic measurements of ocean physics and chemistry were tabulated. </p><p> <b>Chapter 6.</b> We report on the microbial diversity across the Indian Ocean and a lagoon in the Chagos Archipelago. The community within the lagoon differed from adjacent community despite constant water exchange, driven by photosynthetic cyanobacterium Synechococcus. Enrichment of photosynthesis-related transcripts and nutrient cycling indicate influence of primary production on community structure. A five-fold diurnal increase in viral transcripts within the lagoon suggests concomitant bacteriophage influence. </p><p> <b>Chapter 7.</b> We present a novel ChIP-seq workflow for archaea using <i>Halobacterium salinarum</i> sp. NRC-1 and map binding sites of natively expressed transcription factors. Relative to ChIP-Chip and qPCR, it improves spatial resolution and reduces cost. </p><p> <b>Chapter 8.</b> Pique is a user-friendly, freely licensed ChIP-Seq peak finding application for bacterial and archaeal ChIP-Seq experiments. Output is provided in standardized file formats for manual curation and data exploration. </p><p> <b>Chapter 9.</b> With appropriate handling, 3D printers produce sterile components from non-sterile thermoplastic feedstock without post-fabrication treatment. </p><p> <b>Chapter 10.</b> We present a method for fabricating single-use microtiter plates with volumes calibrated for each sample, allowing the use of multichannel pipettes for general liquid handling operations. Many custom plates can be 3D printed simultaneously, resulting in substantial savings in cost and time. </p><p> <b>Chapter 11.</b> The growth kinetics of 48 strains of building-associated bacteria were measured aboard the International Space Station. One strain, <i> Bacillus safensis</i> JPL-MERTA-8-2, grew 60% better in microgravity. </p><p>

Identiferoai:union.ndltd.org:PROQUEST/oai:pqdtoai.proquest.com:10686826
Date01 June 2018
CreatorsNeches, Russell Y.
PublisherUniversity of California, Davis
Source SetsProQuest.com
LanguageEnglish
Detected LanguageEnglish
Typethesis

Page generated in 0.0017 seconds