A fast comparative genome browser for diverse bacteria and archaea.

Genome sequencing has revealed an incredible diversity of bacteria and archaea, but there are no fast and convenient tools for browsing across these genomes. It is cumbersome to view the prevalence of homologs for a protein of interest, or the gene neighborhoods of those homologs, across the diversity of the prokaryotes. We developed a web-based tool, fast.genomics, that uses two strategies to support fast browsing across the diversity of prokaryotes. First, the database of genomes is split up. The main database contains one representative from each of the 6,377 genera that have a high-quality genome, and additional databases for each taxonomic order contain up to 10 representatives of each species. Second, homologs of proteins of interest are identified quickly by using accelerated searches, usually in a few seconds. Once homologs are identified, fast.genomics can quickly show their prevalence across taxa, view their neighboring genes, or compare the prevalence of two different proteins. Fast.genomics is available at https://fast.genomics.lbl.gov.

Taxonomic and environmental distribution of bacterial amino acid auxotrophies.

Many microorganisms are auxotrophic-unable to synthesize the compounds they require for growth. With this work, we quantify the prevalence of amino acid auxotrophies across a broad diversity of bacteria and habitats. We predicted the amino acid biosynthetic capabilities of 26,277 unique bacterial genomes spanning 12 phyla using a metabolic pathway model validated with empirical data. Amino acid auxotrophy is widespread across bacterial phyla, but we conservatively estimate that the majority of taxa (78.4%) are able to synthesize all amino acids. Our estimates indicate that amino acid auxotrophies are more prevalent among obligate intracellular parasites and in free-living taxa with genomic attributes characteristic of 'streamlined' life history strategies. We predicted the amino acid biosynthetic capabilities of bacterial communities found in 12 unique habitats to investigate environmental associations with auxotrophy, using data compiled from 3813 samples spanning major aquatic, terrestrial, and engineered environments. Auxotrophic taxa were more abundant in host-associated environments (including the human oral cavity and gut) and in fermented food products, with auxotrophic taxa being relatively rare in soil and aquatic systems. Overall, this work contributes to a more complete understanding of amino acid auxotrophy across the bacterial tree of life and the ecological contexts in which auxotrophy can be a successful strategy.

Evaluating E. coli genome-scale metabolic model accuracy with high-throughput mutant fitness data.

The Escherichia coli genome-scale metabolic model (GEM) is an exemplar systems biology model for the simulation of cellular metabolism. Experimental validation of model predictions is essential to pinpoint uncertainty and ensure continued development of accurate models. Here, we quantified the accuracy of four subsequent E. coli GEMs using published mutant fitness data across thousands of genes and 25 different carbon sources. This evaluation demonstrated the utility of the area under a precision-recall curve relative to alternative accuracy metrics. An analysis of errors in the latest (iML1515) model identified several vitamins/cofactors that are likely available to mutants despite being absent from the experimental growth medium and highlighted isoenzyme gene-protein-reaction mapping as a key source of inaccurate predictions. A machine learning approach further identified metabolic fluxes through hydrogen ion exchange and specific central metabolism branch points as important determinants of model accuracy. This work outlines improved practices for the assessment of GEM accuracy with high-throughput mutant fitness data and highlights promising areas for future model refinement in E. coli and beyond.

Large-scale genetic characterization of the model sulfate-reducing bacterium, Desulfovibrio vulgaris Hildenborough.

Sulfate-reducing bacteria (SRB) are obligate anaerobes that can couple their growth to the reduction of sulfate. Despite the importance of SRB to global nutrient cycles and their damage to the petroleum industry, our molecular understanding of their physiology remains limited. To systematically provide new insights into SRB biology, we generated a randomly barcoded transposon mutant library in the model SRB Desulfovibrio vulgaris Hildenborough (DvH) and used this genome-wide resource to assay the importance of its genes under a range of metabolic and stress conditions. In addition to defining the essential gene set of DvH, we identified a conditional phenotype for 1,137 non-essential genes. Through examination of these conditional phenotypes, we were able to make a number of novel insights into our molecular understanding of DvH, including how this bacterium synthesizes vitamins. For example, we identified DVU0867 as an atypical L-aspartate decarboxylase required for the synthesis of pantothenic acid, provided the first experimental evidence that biotin synthesis in DvH occurs via a specialized acyl carrier protein and without methyl esters, and demonstrated that the uncharacterized dehydrogenase DVU0826:DVU0827 is necessary for the synthesis of pyridoxal phosphate. In addition, we used the mutant fitness data to identify genes involved in the assimilation of diverse nitrogen sources and gained insights into the mechanism of inhibition of chlorate and molybdate. Our large-scale fitness dataset and RB-TnSeq mutant library are community-wide resources that can be used to generate further testable hypotheses into the gene functions of this environmentally and industrially important group of bacteria.

Interactive Analysis of Functional Residues in Protein Families.

A protein's function depends on functional residues that determine its binding specificity or its catalytic activity, but these residues are typically not considered when annotating a protein's function. To help biologists investigate the functional residues of proteins, we developed two interactive web-based tools, SitesBLAST and Sites on a Tree. Given a protein sequence, SitesBLAST finds homologs that have known functional residues and shows whether the functional residues are conserved. Sites on a Tree shows how functional residues vary across a protein family by showing them on a phylogenetic tree. These tools are available at http://papers.genomics.lbl.gov/sites. IMPORTANCE For most microbes of interest, a genome sequence is available, but the function of its proteins is not known. Instead, proteins' functions are predicted from their similarity to other protein sequences. Within a protein's sequence, a few key residues are most important for function, such as catalyzing a chemical reaction or determining what it binds. But most function prediction tools do not take these key residues into account. We developed interactive tools for identifying functional residues in a protein sequence by comparing it to proteins with known functional residues. Our tools also make it easy to compare key residues across many similar proteins. This should help biologists check if a protein's function is predicted correctly, or to predict if groups of similar proteins have conserved functions.

Filling gaps in bacterial catabolic pathways with computation and high-throughput genetics.

To discover novel catabolic enzymes and transporters, we combined high-throughput genetic data from 29 bacteria with an automated tool to find gaps in their catabolic pathways. GapMind for carbon sources automatically annotates the uptake and catabolism of 62 compounds in bacterial and archaeal genomes. For the compounds that are utilized by the 29 bacteria, we systematically examined the gaps in GapMind's predicted pathways, and we used the mutant fitness data to find additional genes that were involved in their utilization. We identified novel pathways or enzymes for the utilization of glucosamine, citrulline, myo-inositol, lactose, and phenylacetate, and we annotated 299 diverged enzymes and transporters. We also curated 125 proteins from published reports. For the 29 bacteria with genetic data, GapMind finds high-confidence paths for 85% of utilized carbon sources. In diverse bacteria and archaea, 38% of utilized carbon sources have high-confidence paths, which was improved from 27% by incorporating the fitness-based annotations and our curation. GapMind for carbon sources is available as a web server (http://papers.genomics.lbl.gov/carbon) and takes just 30 seconds for the typical genome.

Systematic discovery of pseudomonad genetic factors involved in sensitivity to tailocins.

Tailocins are bactericidal protein complexes produced by a wide variety of bacteria that kill closely related strains and may play a role in microbial community structure. Thanks to their high specificity, tailocins have been proposed as precision antibacterial agents for therapeutic applications. Compared to tailed phages, with whom they share an evolutionary and morphological relationship, bacterially produced tailocins kill their host upon production but producing strains display resistance to self-intoxication. Though lipopolysaccharide (LPS) has been shown to act as a receptor for tailocins, the breadth of factors involved in tailocin sensitivity, and the mechanisms behind resistance to self-intoxication, remain unclear. Here, we employed genome-wide screens in four non-model pseudomonads to identify mutants with altered fitness in the presence of tailocins produced by closely related pseudomonads. Our mutant screens identified O-antigen composition and display as most important in defining sensitivity to our tailocins. In addition, the screens suggest LPS thinning as a mechanism by which resistant strains can become more sensitive to tailocins. We validate many of these novel findings, and extend these observations of tailocin sensitivity to 130 genome-sequenced pseudomonads. This work offers insights into tailocin-bacteria interactions, informing the potential use of tailocins in microbiome manipulation and antibacterial therapy.

Functional genetics of human gut commensal Bacteroides thetaiotaomicron reveals metabolic requirements for growth across environments.

Harnessing the microbiota for beneficial outcomes is limited by our poor understanding of the constituent bacteria, as the functions of most of their genes are unknown. Here, we measure the growth of a barcoded transposon mutant library of the gut commensal Bacteroides thetaiotaomicron on 48 carbon sources, in the presence of 56 stress-inducing compounds, and during mono-colonization of gnotobiotic mice. We identify 516 genes with a specific phenotype under only one or a few conditions, enabling informed predictions of gene function. For example, we identify a glycoside hydrolase important for growth on type I rhamnogalacturonan, a DUF4861 protein for glycosaminoglycan utilization, a 3-keto-glucoside hydrolase for disaccharide utilization, and a tripartite multidrug resistance system specifically for bile salt tolerance. Furthermore, we show that B. thetaiotaomicron uses alternative enzymes for synthesizing nitrogen-containing metabolic precursors based on ammonium availability and that these enzymes are used differentially in vivo in a diet-dependent manner.

Four families of folate-independent methionine synthases.

Although most organisms synthesize methionine from homocysteine and methyl folates, some have "core" methionine synthases that lack folate-binding domains and use other methyl donors. In vitro, the characterized core synthases use methylcobalamin as a methyl donor, but in vivo, they probably rely on corrinoid (vitamin B12-binding) proteins. We identified four families of core methionine synthases that are distantly related to each other (under 30% pairwise amino acid identity). From the characterized enzymes, we identified the families MesA, which is found in methanogens, and MesB, which is found in anaerobic bacteria and archaea with the Wood-Ljungdahl pathway. A third uncharacterized family, MesC, is found in anaerobic archaea that have the Wood-Ljungdahl pathway and lack known forms of methionine synthase. We predict that most members of the MesB and MesC families accept methyl groups from the iron-sulfur corrinoid protein of that pathway. The fourth family, MesD, is found only in aerobic bacteria. Using transposon mutants and complementation, we show that MesD does not require 5-methyltetrahydrofolate or cobalamin. Instead, MesD requires an uncharacterized protein family (DUF1852) and oxygen for activity.

High-throughput mapping of the phage resistance landscape in E. coli.

Bacteriophages (phages) are critical players in the dynamics and function of microbial communities and drive processes as diverse as global biogeochemical cycles and human health. Phages tend to be predators finely tuned to attack specific hosts, even down to the strain level, which in turn defend themselves using an array of mechanisms. However, to date, efforts to rapidly and comprehensively identify bacterial host factors important in phage infection and resistance have yet to be fully realized. Here, we globally map the host genetic determinants involved in resistance to 14 phylogenetically diverse double-stranded DNA phages using two model Escherichia coli strains (K-12 and BL21) with known sequence divergence to demonstrate strain-specific differences. Using genome-wide loss-of-function and gain-of-function genetic technologies, we are able to confirm previously described phage receptors as well as uncover a number of previously unknown host factors that confer resistance to one or more of these phages. We uncover differences in resistance factors that strongly align with the susceptibility of K-12 and BL21 to specific phage. We also identify both phage-specific mechanisms, such as the unexpected role of cyclic-di-GMP in host sensitivity to phage N4, and more generic defenses, such as the overproduction of colanic acid capsular polysaccharide that defends against a wide array of phages. Our results indicate that host responses to phages can occur via diverse cellular mechanisms. Our systematic and high-throughput genetic workflow to characterize phage-host interaction determinants can be extended to diverse bacteria to generate datasets that allow predictive models of how phage-mediated selection will shape bacterial phenotype and evolution. The results of this study and future efforts to map the phage resistance landscape will lead to new insights into the coevolution of hosts and their phage, which can ultimately be used to design better phage therapeutic treatments and tools for precision microbiome engineering.

GapMind: Automated Annotation of Amino Acid Biosynthesis.

GapMind is a Web-based tool for annotating amino acid biosynthesis in bacteria and archaea (http://papers.genomics.lbl.gov/gaps). GapMind incorporates many variant pathways and 130 different reactions, and it analyzes a genome in just 15 s. To avoid error-prone transitive annotations, GapMind relies primarily on a database of experimentally characterized proteins. GapMind correctly handles fusion proteins and split proteins, which often cause errors for best-hit approaches. To improve GapMind's coverage, we examined genetic data from 35 bacteria that grow in defined media without amino acids, and we filled many gaps in amino acid biosynthesis pathways. For example, we identified additional genes for arginine synthesis with succinylated intermediates in Bacteroides thetaiotaomicron, and we propose that Dyella japonica synthesizes tyrosine from phenylalanine. Nevertheless, for many bacteria and archaea that grow in minimal media, genes for some steps still cannot be identified. To help interpret potential gaps, GapMind checks if they match known gaps in related microbes that can grow in minimal media. GapMind should aid the identification of microbial growth requirements.IMPORTANCE Many microbes can make all of the amino acids (the building blocks of proteins). In principle, we should be able to predict which amino acids a microbe can make, and which it requires as nutrients, by checking its genome sequence for all of the necessary genes. However, in practice, it is difficult to check for all of the alternative pathways. Furthermore, new pathways and enzymes are still being discovered. We built an automated tool, GapMind, to annotate amino acid biosynthesis in bacterial and archaeal genomes. We used GapMind to list gaps: cases where a microbe makes an amino acid but a complete pathway cannot be identified in its genome. We used these gaps, together with data from mutants, to identify new pathways and enzymes. However, for most bacteria and archaea, we still do not know how they can make all of the amino acids.

Selective carbon sources influence the end products of microbial nitrate respiration.

Respiratory and catabolic genes are differentially distributed across microbial genomes. Thus, specific carbon sources may favor different respiratory processes. We profiled the influence of 94 carbon sources on the end products of nitrate respiration in microbial enrichment cultures from diverse terrestrial environments. We found that some carbon sources consistently favor dissimilatory nitrate reduction to ammonium (DNRA/nitrate ammonification) while other carbon sources favor nitrite accumulation or denitrification. For an enrichment culture from aquatic sediment, we sequenced the genomes of the most abundant strains, matched these genomes to 16S rDNA exact sequence variants (ESVs), and used 16S rDNA amplicon sequencing to track the differential enrichment of functionally distinct ESVs on different carbon sources. We found that changes in the abundances of strains with different genetic potentials for nitrite accumulation, DNRA or denitrification were correlated with the nitrite or ammonium concentrations in the enrichment cultures recovered on different carbon sources. Specifically, we found that either L-sorbose or D-cellobiose enriched for a Klebsiella nitrite accumulator, other sugars enriched for an Escherichia nitrate ammonifier, and citrate or formate enriched for a Pseudomonas denitrifier and a Sulfurospirillum nitrate ammonifier. Our results add important nuance to the current paradigm that higher concentrations of carbon will always favor DNRA over denitrification or nitrite accumulation, and we propose that, in some cases, carbon composition can be as important as carbon concentration in determining nitrate respiratory end products. Furthermore, our approach can be extended to other environments and metabolisms to characterize how selective parameters influence microbial community composition, gene content, and function.

Nitrate-Utilizing Microorganisms Resistant to Multiple Metals from the Heavily Contaminated Oak Ridge Reservation.

Contamination of environments with nitrate generated by industrial processes and the use of nitrogen-containing fertilizers is a growing problem worldwide. While nitrate can be removed from contaminated areas by microbial denitrification, nitrate frequently occurs with other contaminants, such as heavy metals, that have the potential to impede the process. Here, nitrate-reducing microorganisms were enriched and isolated from both groundwater and sediments at the Oak Ridge Reservation (ORR) using concentrations of nitrate and metals (Al, Mn, Fe, Co, Ni, Cu, Cd, and U) similar to those observed in a contaminated environment at ORR. Seven new metal-resistant, nitrate-reducing strains were characterized, and their distribution across both noncontaminated and contaminated areas at ORR was examined. While the seven strains have various pH ranges for growth, carbon source preferences, and degrees of resistance to individual and combinations of metals, all were able to reduce nitrate at similar rates both in the presence and absence of the mixture of metals found in the contaminated ORR environment. Four strains were identified in groundwater samples at different ORR locations by exact 16S RNA sequence variant analysis, and all four were found in both noncontaminated and contaminated areas. By using environmentally relevant metal concentrations, we successfully isolated multiple organisms from both ORR noncontaminated and contaminated environments that are capable of reducing nitrate in the presence of extreme mixed-metal contamination.IMPORTANCE Nitrate contamination is a global issue that affects groundwater quality. In some cases, cocontamination of groundwater with nitrate and mixtures of heavy metals could decrease microbially mediated nitrate removal, thereby increasing the duration of nitrate contamination. Here, we used metal and nitrate concentrations that are present in a contaminated site at the Oak Ridge Reservation to isolate seven metal-resistant strains. All were able to reduce nitrate in the presence of high concentrations of a mixture of heavy metals. Four of seven strains were located in pristine as well as contaminated sites at the Oak Ridge Reservation. Further study of these nitrate-reducing strains will uncover mechanisms of resistance to multiple metals that will increase our understanding of the effect of nitrate and metal contamination on groundwater microbial communities.

Curated BLAST for Genomes.

Curated BLAST for Genomes finds candidate genes for a process or an enzymatic activity within a genome of interest. In contrast to annotation tools, which usually predict a single activity for each protein, Curated BLAST asks if any of the proteins in the genome are similar to characterized proteins that are relevant. Given a query such as an enzyme's name or an EC number, Curated BLAST searches the curated descriptions of over 100,000 characterized proteins, and it compares the relevant characterized proteins to the predicted proteins in the genome of interest. In case of errors in the gene models, Curated BLAST also searches the six-frame translation of the genome. Curated BLAST is available at http://papers.genomics.lbl.gov/curated. IMPORTANCE Given a microbe's genome sequence, we often want to predict what capabilities the organism has, such as which nutrients it requires or which energy sources it can use. Or, we know the organism has a capability and we want to find the genes involved. Scientists often use automated gene annotations to find relevant genes, but automated annotations are often vague or incorrect. Curated BLAST finds candidate genes for a capability without relying on automated annotations. First, Curated BLAST finds proteins (usually from other organisms) whose functions have been studied experimentally and whose curated descriptions match a query. Then, it searches the genome of interest for similar proteins and returns a list of candidates. Curated BLAST is fast and often finds relevant genes that are missed by automated annotation.

Correction: Filling gaps in bacterial amino acid biosynthesis pathways with high-throughput genetics.

[This corrects the article DOI: 10.1371/journal.pgen.1007147.].

Oxidative Pathways of Deoxyribose and Deoxyribonate Catabolism.

Using genome-wide mutant fitness assays in diverse bacteria, we identified novel oxidative pathways for the catabolism of 2-deoxy-d-ribose and 2-deoxy-d-ribonate. We propose that deoxyribose is oxidized to deoxyribonate, oxidized to ketodeoxyribonate, and cleaved to acetyl coenzyme A (acetyl-CoA) and glyceryl-CoA. We have genetic evidence for this pathway in three genera of bacteria, and we confirmed the oxidation of deoxyribose to ketodeoxyribonate in vitro. In Pseudomonas simiae, the expression of enzymes in the pathway is induced by deoxyribose or deoxyribonate, while in Paraburkholderia bryophila and in Burkholderia phytofirmans, the pathway proceeds in parallel with the known deoxyribose 5-phosphate aldolase pathway. We identified another oxidative pathway for the catabolism of deoxyribonate, with acyl-CoA intermediates, in Klebsiella michiganensis. Of these four bacteria, only P. simiae relies entirely on an oxidative pathway to consume deoxyribose. The deoxyribose dehydrogenase of P. simiae is either nonspecific or evolved recently, as this enzyme is very similar to a novel vanillin dehydrogenase from Pseudomonas putida that we identified. So, we propose that these oxidative pathways evolved primarily to consume deoxyribonate, which is a waste product of metabolism. IMPORTANCE Deoxyribose is one of the building blocks of DNA and is released when cells die and their DNA degrades. We identified a bacterium that can grow with deoxyribose as its sole source of carbon even though its genome does not contain any of the known genes for breaking down deoxyribose. By growing many mutants of this bacterium together on deoxyribose and using DNA sequencing to measure the change in the mutants' abundance, we identified multiple protein-coding genes that are required for growth on deoxyribose. Based on the similarity of these proteins to enzymes of known function, we propose a 6-step pathway in which deoxyribose is oxidized and then cleaved. Diverse bacteria use a portion of this pathway to break down a related compound, deoxyribonate, which is a waste product of metabolism. Our study illustrates the utility of large-scale bacterial genetics to identify previously unknown metabolic pathways.

Dual-barcoded shotgun expression library sequencing for high-throughput characterization of functional traits in bacteria.

A major challenge in genomics is the knowledge gap between sequence and its encoded function. Gain-of-function methods based on gene overexpression are attractive avenues for phenotype-based functional screens, but are not easily applied in high-throughput across many experimental conditions. Here, we present Dual Barcoded Shotgun Expression Library Sequencing (Dub-seq), a method that uses random DNA barcodes to greatly increase experimental throughput. As a demonstration of this approach, we construct a Dub-seq library with Escherichia coli genomic DNA, performed 155 genome-wide fitness assays in 52 experimental conditions, and identified overexpression phenotypes for 813 genes. We show that Dub-seq data is reproducible, accurately recapitulates known biology, and identifies hundreds of novel gain-of-function phenotypes for E. coli genes, a subset of which we verified with assays of individual strains. Dub-seq provides complementary information to loss-of-function approaches and will facilitate rapid and systematic functional characterization of microbial genomes.

The selective pressures on the microbial community in a metal-contaminated aquifer.

In many environments, toxic compounds restrict which microorganisms persist. However, in complex mixtures of inhibitory compounds, it is challenging to determine which specific compounds cause changes in abundance and prevent some microorganisms from growing. We focused on a contaminated aquifer in Oak Ridge, Tennessee, USA that has large gradients of pH and widely varying concentrations of uranium, nitrate, and many other inorganic ions. In the most contaminated wells, the microbial community is enriched in the Rhodanobacter genus. Rhodanobacter abundance is positively correlated with low pH and high concentrations of uranium and 13 other ions and we sought to determine which of these ions are selective pressures that favor the growth of Rhodanobacter over other taxa. Of these ions, low pH and high UO22+, Mn2+, Al3+, Cd2+, Zn2+, Co2+, and Ni2+ are both (a) selectively inhibitory of a Pseudomonas isolate from an uncontaminated well vs. a Rhodanobacter isolate from a contaminated well, and (b) reach toxic concentrations (for the Pseudomonas isolate) in the Rhodanobacter-dominated wells. We used mixtures of ions to simulate the groundwater conditions in the most contaminated wells and verified that few isolates aside from Rhodanobacter can tolerate these eight ions. These results clarify which ions are likely causal factors that impact the microbial community at this field site and are not merely correlated with taxonomic shifts. Furthermore, our general high-throughput approach can be applied to other environments, isolates, and conditions to systematically help identify selective pressures on microbial communities.

Mutant phenotypes for thousands of bacterial genes of unknown function.

One-third of all protein-coding genes from bacterial genomes cannot be annotated with a function. Here, to investigate the functions of these genes, we present genome-wide mutant fitness data from 32 diverse bacteria across dozens of growth conditions. We identified mutant phenotypes for 11,779 protein-coding genes that had not been annotated with a specific function. Many genes could be associated with a specific condition because the gene affected fitness only in that condition, or with another gene in the same bacterium because they had similar mutant phenotypes. Of the poorly annotated genes, 2,316 had associations that have high confidence because they are conserved in other bacteria. By combining these conserved associations with comparative genomics, we identified putative DNA repair proteins; in addition, we propose specific functions for poorly annotated enzymes and transporters and for uncharacterized protein families. Our study demonstrates the scalability of microbial genetics and its utility for improving gene annotations.

Filling gaps in bacterial amino acid biosynthesis pathways with high-throughput genetics.

For many bacteria with sequenced genomes, we do not understand how they synthesize some amino acids. This makes it challenging to reconstruct their metabolism, and has led to speculation that bacteria might be cross-feeding amino acids. We studied heterotrophic bacteria from 10 different genera that grow without added amino acids even though an automated tool predicts that the bacteria have gaps in their amino acid synthesis pathways. Across these bacteria, there were 11 gaps in their amino acid biosynthesis pathways that we could not fill using current knowledge. Using genome-wide mutant fitness data, we identified novel enzymes that fill 9 of the 11 gaps and hence explain the biosynthesis of methionine, threonine, serine, or histidine by bacteria from six genera. We also found that the sulfate-reducing bacterium Desulfovibrio vulgaris synthesizes homocysteine (which is a precursor to methionine) by using DUF39, NIL/ferredoxin, and COG2122 proteins, and that homoserine is not an intermediate in this pathway. Our results suggest that most free-living bacteria can likely make all 20 amino acids and illustrate how high-throughput genetics can uncover previously-unknown amino acid biosynthesis genes.