A Food Behavior Checklist (FBC) was developed, validated, and pilot-tested with homemakers in Virginia's Expanded Food and Nutrition Education Program (EFNEP). The FBC was designed to measure nutrition-related practices that are taught in EFNEP, but cannot be measured by the 24-hour recall. During development, 20 EFNEP paraprofessionals in Virginia and 20 randomly selected state EFNEP coordinators helped to identify items that are most important in evaluating the overall effectiveness of EFNEP, and which could be used to establish content, face, and construct validity of a behavior assessment instrument. The Food Behavior Checklist contained 30 items, and responses were recorded on a Likert scale with four response levels: 0 = "never or seldom", 1 = "sometimes",2 = "usually", and 3 = "almost always". Cronbach-alpha revealed a reliability coefficient of 0.86. During the field test phase of this study, the FBC was used before and after EFNEP intervention to collect data on the nutrition-related practices of 147 low-income homemakers in three rural counties and one urban area in Virginia. In this phase, the experimental group, which consisted of 79 homemakers, received computer-aided feedback on their dietary practices, via a Diagnostic Report, which was used as a teaching tool. Paraprofessionals thoroughly explained the content of the diagnostic report to the homemaker at program entry and program exit. The control group, which consisted of 68 homemakers, did not receive computer-aided feedback on their dietary intakes, in that no mention was made of the computer print out to them. Results indicated that this instrument (FBC) may be useful in evaluating the overall effectiveness of EFNEP nationwide and may be useful in other nutrition programs. No significant differences were observed in the nutrient intake or Food Behavior Checklist practice change scores between the group who received computer-aided feedback and the group who did not received computer-aided feedback. The author concluded that the dietary analysis contained in the computer generated Diagnostic Report needs to be simplified if it is to become an effective tool with EFNEP homemakers. More research is needed on the use of computerized diet analysis as a teaching tool with low-income homemakers. / Ph. D.

Syftet med litteraturstudien är att sammanfatta vilken gensekvens som används vid genetisk barcoding av växter och hur väl metoden i fråga tillämpas i tre biologiska yrkesområden: dietanalyser i ekologin, analys av pollensporer i forensisk biologi samt analys av uråldrigt DNA (ancient DNA) i paleontologin. Vidare var det även av intresse att se hur genetisk barcoding kan användas i den gymnasiala undervisningen och hur väl den passar in med de svenska styrdokumenten för skolan. Hur elever har gynnats av den valda metoden samt vilka begränsningar som har uppstått har också berörts. Litteraturstudien baseras på vetenskapliga artiklar som har sökts fram med de nedan listade nyckelorden. Resultaten visar att en kombination av gensekvenser, däribland rbcL, matK, trnH-psbA och ITS, fungerar bäst vid identifiering av växter. I dagsläget är genetisk barcoding fortfarande i utvecklingsfasen, där metoden begränsas av antalet referenssekvenser i databaserna, vilket gör det svårt att utesluta morfologiska identifieringsmetoder i de tre yrkesområdena. Vid användning av barcoding i den gymnasiala undervisningen visar det sig att det stämmer väl överens med de svenska styrdokumenten och ökar elevers intresse för de naturvetenskapliga ämnena, då de kan bidra med värdefull forskning genom tillägg av referenssekvenser i databaserna. De största begränsningarna är att det blir ett stort arbetslass för läraren, att läraren i fråga måste vara bekväm med de olika laboratiska momenten och att skolan ska ha tillgång till nödvändig apparatur. / The purpose of the literature study is to conclude which gene sequences are being used in DNA barcoding on plants and how the method in question is being used in three different biological occupations: diet analysis in ecology, analysis of pollen in forensics and analysis of ancient DNA (aDNA) in paleontology. Further it was also of interest to study how DNA barcoding can be used in high school settings and how the method correlates with the Swedish curriculum. How pupils have benefited from the chosen method and what limitations have arisen have also been touched upon. This literature study is based on scientific articles that have been sought with the keywords listed below. The results show that a combination of gene sequences, including rbcL, matK, trnH-psbA and ITS, works best in plant identification. At present, genetic barcoding is still in the developmental phase, where the method is limited by the number of reference sequences in the databases, which makes it difficult to exclude morphological-based methods in the three occupational fields. When using barcoding in upper secondary education it turns out that it’s in good agreement with the Swedish curriculum and increases the students' interest in the scientific subjects, since they can contribute with genuine research when adding reference sequences in the databases. The main limitations are the workload for the teacher, the teacher in question must be comfortable with the different laboratory steps and that the school must have access to necessary equipment.

Cette thèse s'intéresse à la conception et le développement des techniques de bioinfor- matique qui peuvent faciliter l'utilisation de l'approche metabarcoding pour mesurer la diversité d'espèces. Le metabarcoding peut être utilisé avec le séquencage haut débit pour l'identification d'espèces multiples à partir d'un seul échantillon environnemental. La véritable force du metabarcoding réside dans l'utilisation de barcode marqueurs choisi pour une étude particulière et l'identification d'espèces ou des taxons peut être réalisé avec des marqueurs soigneusement conçu. Avec l'avancement des techniques haut débit de séquençage, une énorme quantité des données de séquences est produit qui contient un nombres substantiel des mutations. Ces mutations posent un grand problème pour les estimations correctes de la biodiversité et pour le d'assignation de taxon. Les trois problèmes majeurs dans le domaine de la bioinformatique que j'ai abordés dans cette thèse sont: i) évaluer la qualité d'une barcode marker , ii) concevoir des nouveaux région barcode et iii) d'analyser les données de séquençage pour traiter les erreurs et éliminer le bruit en séquences. Pour évaluer la qualité d'un barcode marker, on a développé deux mesures quantita- tive,formelle: la couverture (Bc) et la spécificité (Bs). La couverture donne une mesure de universalité d'une pairs de primer pour amplifier un large nombre de taxa, alors que la spécificité donne une mesure de capacité à discriminer entre les différents taxons. Ces mesures sont très utiles pour le classement des barcode marker et pour sélectionner les meilleurs markers. Pour trouver des nouveaux région barcode notamment pour les applications metabarcod- ing, j'ai développé un logiciel, ecoPrimers3. Basé sur ces deux mesures de qualité et de l'information taxinomique intégré, ecoPrimers nous permet de concevoir barcode markers pour n'importe quel niveau taxonomique . En plus, avec un grand nombre de paramètres réglables il nous permet de contrôler les propriétés des amorces. Enfin, grâce a des algorithmes efficaces et programmé en langage C, ecoPrimers est suffisamment efficace pour traiter des grosses bases de données, y compris génomes bactériens entièrement séquencés. Enfin pour traiter des erreurs présentes dans les données de séquencage , nous avons analysé un ensemble simple d'échantillons de PCR obtenus à partir de l'analyse du régime alimentaire de Snow Leopard. En mesurant les corrélations entre les différents paramètres des erreurs, nous avons observé que la plupart des erreurs sont produites pendant l'amplification par PCR. Pour détecter ces erreurs, nous avons développé un algorithme utilisant les graphes, qui peuvent différencier les vrai séquences des erreurs induites par PCR. Les résultats obtenus à partir de cet algorithme a montré que les données de-bruitée a donnent une estimation réaliste de la diversité des espèces étudiées dans les Alpes françaises. / This thesis is concerned with the design and development of bioinformatics techniques that can facilitate the use of metabarcoding approach for measuring species diversity. Metabarcoding coupled with next generation sequencing techniques have a strong po- tential for multiple species identification from a single environmental sample. The real strength of metabarcoding resides in the use of barcode markers chosen for a particular study. The identification at species or higher level taxa can be achieved with carefully designed barcode markers. Moreover with the advent of high throughput sequencing techniques huge amount of sequence data is being produced that contains a substantial level of mutations. These mutations pose a problem for the correct estimates of biodi- versity and for the taxon assignation process. Thus the three major challenges that we addressed in this thesis are: evaluating the quality of a barcode region, designing new barcodes and dealing with errors occurring during different steps of an experiment. To assess the quality of a barcode region we have developed two formal quantitative mea- sures called barcode coverage (Bc) and barcode specificity (Bs). Barcode coverage is concerned with the property of a barcode to amplify a broad range of taxa, whereas barcode specificity deals with its ability to discriminate between different taxa. These measures are extremely useful especially for ranking different barcodes and selecting the best markers. To deal with the challenge of designing new barcodes for metabarcoding applications we have developed an efficient software called ecoPrimers. Based on the above two quality measures and with integrated taxonomic information, ecoPrimers1 enables us to design primers and their corresponding barcode markers for any taxonomic level. Moreover with a large number of tunable parameters it allows us to control the properties of primers. Finally, based on efficient algorithms and implemented in C language, ecoPrimers is efficient enough to deal with large data bases including fully sequenced bacterial genomes. Finally to deal with errors present in DNA sequence data, we have analyzed a simple set of PCR samples obtained from the diet analysis of snow leopard. We grouped closely related sequences and by measuring the correlation between different parameters of mutations, we have shown that most of the errors were introduced during PCR amplification. In order to deal with such errors, we have further developed an algorithm using graphs approach, that can differentiate true sequences from PCR induced errors. The results obtained from this algorithm showed that de-noised data gave a realistic estimate of species diversity studied in French Alpes. This algorithm is implemented in program obiclean.