A field guide to cultivating computational biology

Evolving in sync with the computation revolution over the past 30 years, computational biology has emerged as a mature scientific field. While the field has made major contributions toward improving scientific knowledge and human health, individual computational biology practitioners at various institutions often languish in career development. As optimistic biologists passionate about the future of our field, we propose solutions for both eager and reluctant individual scientists, institutions, publishers, funding agencies, and educators to fully embrace computational biology. We believe that in order to pave the way for the next generation of discoveries, we need to improve recognition for computational biologists and better align pathways of career success with pathways of scientific progress. With 10 outlined steps, we call on all adjacent fields to move away from the traditional individual, single-discipline investigator research model and embrace multidisciplinary, data-driven, team science.

Do you want to attract computational biologists to your project or to your department? Despite the major contributions of computational biology, those attempting to bridge the interdisciplinary gap often languish in career advancement, publication, and grant review. Here, sixteen computational biologists around the globe present "A field guide to cultivating computational biology," focusing on solutions.

Biology in the digital era requires computation and collaboration. A modern research project may include multiple model systems, use multiple assay technologies, collect varying data types, and require complex computational strategies, which together make effective design and execution difficult or impossible for any individual scientist. While some labs, institutions, funding bodies, publishers, and other educators have already embraced a team science model in computational biology and thrived [1–7], others who have not yet fully adopted it risk severely lagging behind the cutting edge. We propose a general solution: “deep integration” between biology and the computational sciences. Many different collaborative models can yield deep integration, and different problems require different approaches (Fig 1).Open in a separate windowFig 1Supporting interdisciplinary team science will accelerate biological discoveries.Scientists who have little exposure to different fields build silos, in which they perform science without external input. To solve hard problems and to extend your impact, collaborate with diverse scientists, communicate effectively, recognize the importance of core facilities, and embrace research parasitism. In biologically focused parasitism, wet lab biologists use existing computational tools to solve problems; in computationally focused parasitism, primarily dry lab biologists analyze publicly available data. Both strategies maximize the use and societal benefit of scientific data.In this article, we define computational science extremely broadly to include all quantitative approaches such as computer science, statistics, machine learning, and mathematics. We also define biology broadly, including any scientific inquiry pertaining to life and its many complications. A harmonious deep integration between biology and computer science requires action—we outline 10 immediate calls to action in this article and aim our speech directly at individual scientists, institutions, funding agencies, and publishers in an attempt to shift perspectives and enable action toward accepting and embracing computational biology as a mature, necessary, and inevitable discipline (Box 1).Box 1. Ten calls to action for individual scientists, funding bodies, publishers, and institutions to cultivate computational biology. Many actions require increased funding support, while others require a perspective shift. For those actions that require funding, we believe convincing the community of need is the first step toward agencies and systems allocating sufficient support

1. Respect collaborators’ specific research interests and motivationsProblem: Researchers face conflicts when their goals do not align with collaborators. For example, projects with routine analyses provide little benefit for computational biologists.Solution: Explicit discussion about interests/expertise/goals at project onset.Opportunity: Clearly defined expectations identify gaps, provide commitment to mutual benefit.
2. Seek necessary input during project design and throughout the project life cycleProblem: Modern research projects require multiple experts spanning the project’s complexity.Solution: Engage complementary scientists with necessary expertise throughout the entire project life cycle.Opportunity: Better designed and controlled studies with higher likelihood for success.
3. Provide and preserve budgets for computational biologists’ workProblem: The perception that analysis is “free” leads to collaborator budget cuts.Solution: When budget cuts are necessary, ensure that they are spread evenly.Opportunity: More accurate, reproducible, and trustworthy computational analyses.
4. Downplay publication author order as an evaluation metric for computational biologistsProblem: Computational biologist roles on publications are poorly understood and undervalued.Solution: Journals provide more equitable opportunities, funding bodies and institutions improve understanding of the importance of team science, scientists educate each other.Opportunity: Engage more computational biologist collaborators, provide opportunities for more high-impact work.
5. Value software as an academic productProblem: Software is relatively undervalued and can end up poorly maintained and supported, wasting the time put into its creation.Solution: Scientists cite software, and funding bodies provide more software funding opportunities.Opportunity: More high-quality maintainable biology software will save time, reduce reimplementation, and increase analysis reproducibility.
6. Establish academic structures and review panels that specifically reward team scienceProblem: Current mechanisms do not consistently reward multidisciplinary work.Solution: Separate evaluation structures to better align peer review to reward indicators of team science.Opportunity: More collaboration to attack complex multidisciplinary problems.
7. Develop and reward cross-disciplinary training and mentoringProblem: Academic labs and institutions are often insufficiently equipped to provide training to tackle the next generation of biological problems, which require computational skills.Solution: Create better training programs aligned to necessary on-the-job skills with an emphasis on communication, encourage wet/dry co-mentorship, and engage younger students to pursue computational biology.Opportunity: Interdisciplinary students uncover important insights in their own data.
8. Support computing and experimental infrastructure to empower computational biologistsProblem: Individual computational labs often fund suboptimal cluster computing systems and lack access to data generation facilities.Solution: Institutions can support centralized compute and engage core facilities to provide data services.Opportunity: Time and cost savings for often overlooked administrative tasks.
9. Provide incentives and mechanisms to share open data to empower discovery through reanalysisProblem: Data are often siloed and have untapped potential.Solution: Provide institutional data storage with standardized identifiers and provide separate funding mechanisms and publishing venues for data reuse.Opportunity: Foster new breed of researchers, “research parasites,” who will integrate multimodal data and enhance mechanistic insights.
10. Consider infrastructural, ethical, and cultural barriers to clinical data accessProblem: Identifiable health data, which include sensitive information that must be kept hidden, are distributed and disorganized, and thus underutilized.Solution: Leadership must enforce policies to share deidentifiable data with interoperable metadata identifiers.Opportunity: Derive new insights from multimodal data integration and build datasets with increased power to make biological discoveries.

     

Age-Related Differences in the Accuracy of Web Query-Based Predictions of Influenza-Like Illness

BackgroundWeb queries are now widely used for modeling, nowcasting and forecasting influenza-like illness (ILI). However, given that ILI attack rates vary significantly across ages, in terms of both magnitude and timing, little is known about whether the association between ILI morbidity and ILI-related queries is comparable across different age-groups. The present study aimed to investigate features of the association between ILI morbidity and ILI-related query volume from the perspective of age.MethodsSince Google Flu Trends is unavailable in Italy, Google Trends was used to identify entry terms that correlated highly with official ILI surveillance data. All-age and age-class-specific modeling was performed by means of linear models with generalized least-square estimation. Hold-out validation was used to quantify prediction accuracy. For purposes of comparison, predictions generated by exponential smoothing were computed.ResultsFive search terms showed high correlation coefficients of > .6. In comparison with exponential smoothing, the all-age query-based model correctly predicted the peak time and yielded a higher correlation coefficient with observed ILI morbidity (.978 vs. .929). However, query-based prediction of ILI morbidity was associated with a greater error. Age-class-specific query-based models varied significantly in terms of prediction accuracy. In the 0–4 and 25–44-year age-groups, these did well and outperformed exponential smoothing predictions; in the 15–24 and ≥ 65-year age-classes, however, the query-based models were inaccurate and highly overestimated peak height. In all but one age-class, peak timing predicted by the query-based models coincided with observed timing.ConclusionsThe accuracy of web query-based models in predicting ILI morbidity rates could differ among ages. Greater age-specific detail may be useful in flu query-based studies in order to account for age-specific features of the epidemiology of ILI.     

Strategies for genetic conservation of trees in the Peruvian Amazon

Forestry practices and high rates of land clearance for agriculture are causing genetic erosion of valuable tree species in the Peruvian Amazon, thereby endangering the economic sustainability of rural communities and limiting Peru's opportunities for the development of new timber and non-timber forest products. The potential utility and limitations of six low-input interventions to help forestall further genetic erosion in the region are discussed, with a focus on local community involvement. Improved agroforestry systems may help reduce deforestation by increasing farm productivity, although methods to increase the currently low adoption rate of these technologies need to be developed. Use of strategic tree domestication techniques can also improve farm productivity and prevent inadvertent genetic drift and inbreeding associated with traditional domestication practices, although to have a major impact, current programs need to be extended across the region. Woodlot forestry could supplant selective extraction of timber and offers an attractive opportunity for poverty alleviation if appropriate credit and land tenure policies can be developed. However, it may also result in increased deforestation if activities on public land cannot be controlled. The implementation of improved seed collection systems and simple seed transfer guidelines would help to reduce the collection of seed of poor quality and low genetic diversity, and avoid maladapted plantings, although such programs are difficult to monitor and seed costs may increase. Strategic identification and design of in situ conservation areas would help to ensure the viability of conserved populations, but requires the forfeiture of significant revenue from timber concessions.     

Sponge-associated microbial Antarctic communities exhibiting antimicrobial activity against Burkholderia cepacia complex bacteria

The aerobic heterotrophic bacterial communities isolated from three different Antarctic sponge species were analyzed for their ability to produce antimicrobial compounds active toward Cystic Fibrosis opportunistic pathogens belonging to the Burkholderia cepacia complex (Bcc). The phylogenetic analysis performed on the 16S rRNA genes affiliated the 140 bacterial strains analyzed to 15 genera. Just three of them (Psychrobacter, Pseudoalteromonas and Arthrobacter) were shared by the three sponges. The further Random Amplified Polymorphic DNA analysis allowed to demonstrate that microbial communities are highly sponge-specific and a very low degree of genus/species/strain sharing was detected. Data obtained revealed that most of these sponge-associated Antarctic bacteria and belonging to different genera were able to completely inhibit the growth of bacteria belonging to the Bcc. On the other hand, the same Antarctic strains did not have any effect on the growth of other pathogenic bacteria, strongly suggesting that the inhibition is specific for Bcc bacteria. Moreover, the antimicrobial compounds synthesized by the most active Antarctic bacteria are very likely Volatile Organic Compounds (VOCs), a finding that was confirmed by the SPME-GC-MS technique, which revealed the production of a large set of VOCs by a representative set of Antarctic bacteria. The synthesis of these VOCs appeared to be related neither to the presence of pks genes nor the presence of plasmid molecules. The whole body of data obtained in this work indicates that sponge-associated bacteria represent an untapped source for the identification of new antimicrobial compounds and are paving the way for the discovery of new drugs that can be efficiently and successfully used for the treatment of CF infections.     

High expression level of a gene coding for a chloroplastic amino acid selective channel protein is correlated to cold acclimation in cereals

A cold-regulated gene (cor tmc-ap3) coding for a putative chloroplastic amino acid selective channel protein was isolated from cold-treated barley leaves combining the differential display and the 5-RACE techniques. Cor tmc-ap3 is expressed at low level under normal growing temperature, and its expression is strongly enhanced after cold treatment. A positive correlation between the expression of cor tmc-ap3 and frost tolerance was found both among barley cultivars and among cereal species. The COR TMC-AP3 protein was expressed in vitro, purified and used to raise a polyclonal antibody. Western analysis showed that the cor tmc-ap3 gene product is localized to the chloroplastic outer envelope fraction, supporting its putative function. The frost-resistant winter cultivar Onice accumulated COR TMC-AP3 more rapidly and at a higher level than the frost-susceptible spring cultivar Gitane. After 28 days of cold acclimation the winter cultivar had about 2-fold more protein than the spring genotype. All these results suggest that an increased amount of a chloroplastic amino acid selective channel protein could be required for cold acclimation in cereals. Hypotheses about the role of COR TMC-AP3 during the hardening process are discussed.     

Fluorescent, internally quenched, peptides for exploring the pH-dependent substrate specificity of cathepsin B.

Cathepsin B is a cysteine protease that in tumor tissues is localized in both acidic lysosomes and extracellular spaces. It can catalyze the cleavage of peptide bonds by two mechanisms: endoproteolytic attack with a pH optimum around 7.4, and attack from the C-terminus with a pH optimum at 4.5-5.5. In this work, seven fluorescent, internally quenched, decapeptides have been synthesized using the prototypical cathepsin B selective substrate Z-Phe-Arg-AMC as a lead, and used to identify the structural factors determining the susceptibility of peptides to hydrolysis at acidic and neutral pH values. Each peptide differs from the others in one amino acid (residue 6) and contains a highly fluorescent Nma group linked to the alpha-amino function of the N-terminal Orn residue and a Dnp group linked to the side chain of the Lys(8) residue acting as a quencher. Proteolytic cleavage was monitored by measuring the increase of fluorescence at 440 nm upon excitation at 340 nm, and the cleavage sites were determined by HPLC followed by ESI-MS analysis. Peptides containing Ala or Phe at position 6 are good substrates for the enzyme at both pH 5.0 and 7.4. By contrast, those containing Glu, Asp, Lys or Val are not cleaved at all by cathepsin B at pH 7.4, and are poorly hydrolyzed at pH 5.0. These findings provide new information for the rational design of cathepsin B-activated peptide-containing anticancer drugs.     

Perturbation of cytochrome P450, generation of oxidative stress and induction of DNA damage in Cyprinus carpio exposed in situ to potable surface water

Epidemiological evidence suggests a link between consumption of chlorinated drinking water and various cancers. Chlorination of water rich in organic chemicals produces carcinogenic organochlorine by-products (OBPs) such as trihalomethanes and haloacetic acids. Since the discovery of the first OBP in the 1970s, there have been several investigations designed to determine the biological effects of single chemicals or small artificial OBP combinations. However, there is still insufficient information regarding the general biological response to these compounds, and further studies are still needed to evaluate their potential genotoxic effects. In the current study, we evaluated the effect of three drinking water disinfectants on the activity of cytochrome P450 (CYP)-linked metabolizing enzymes and on the generation of oxidative stress in the livers of male and female Cyprinus carpio fish (carp). The fish were exposed in situ for up 20 days to surface water obtained from the Trasmene lake in Italy. The water was treated with 1-2 mg/L of either sodium hypochlorite (NaClO) or chlorine dioxide (ClO2) as traditional disinfectants or with a relatively new disinfectant product, peracetic acid (PAA). Micronucleus (MN) frequencies in circulating erythrocytes from the fish were also analysed as a biomarker of genotoxic effect. In the CYP-linked enzyme assays, a significant induction (up to a 57-fold increase in the deethylation of ethoxyresorufin with PAA treatment) and a notable inactivation (up to almost a 90% loss in hydroxylation of p-nitrophenol with all disinfectants, and of testosterone 2beta-hydroxylation with NaClO) was observed in subcellular liver preparations from exposed fish. Using the electron paramagnetic resonance (EPR) spectroscopy radical-probe technique, we also observed that CYP-modulation was associated with the production of reactive oxygen species (ROS). In addition, we found a significant increase in MN frequency in circulating erythrocytes after 10 days of exposure of fish to water treated with ClO2, while a non-significant six-fold increase in MN frequency was observed with NaClO, but not with PAA. Our data suggest that the use of ClO2 and NaClO to disinfect drinking water could generate harmful OBP mixtures that are able to perturb CYP-mediated reactions, generate oxidative stress and induce genetic damage. These data may provide a mechanistic explanation for epidemiological studies linking consumption of chlorinated drinking water to increased risk of urinary, gastrointestinal and bladder cancers.     

Gold(III) compounds as anticancer agents: relevance of gold-protein interactions for their mechanism of action

Gold(III) compounds constitute an emerging class of biologically active substances, of special interest as potential anticancer agents. During the past decade a number of structurally diverse gold(III) complexes were reported to be acceptably stable under physiological-like conditions and to manifest very promising cytotoxic effects against selected human tumour cell lines, making them good candidates as anti-tumour drugs. Some representative examples will be described in detail. There is considerable interest in understanding the precise biochemical mechanisms of these novel cytotoxic agents. Based on experimental evidence collected so far we hypothesize that these metallodrugs, at variance with classical platinum(II) drugs, produce in most cases their growth inhibition effects through a variety of "DNA-independent" mechanisms. Notably, strong inhibition of the selenoenzyme thioredoxin reductase and associated disregulation of mitochondrial functions were clearly documented in some selected cases, thus providing a solid biochemical basis for the pronounced proapoptotic effects. These observations led us to investigate in detail the reactions of gold(III) compounds with a few model proteins in order to gain molecular-level information on the possible interaction modes with possible protein targets. Valuable insight on the formation and the nature of gold-protein adducts was gained through ESI MS (electrospray ionization mass spectrometry) and spectrophotometric studies of appropriate model systems as it is exemplified here by the reactions of two representative gold(III) compounds with cytochrome c and ubiquitin. The mechanistic relevance of gold(III)-induced oxidative protein damage and of direct gold coordination to protein sidechains is specifically assessed. Perspectives for the future of this topics are briefly outlined.