The chromosomes involved in twenty reciprocal translocations in G. hirsutum were reidentified by means of a specially designed crossing program. Each translocation line was crossed to appropriate monosomic, monotelodisomic, and/or translocation lines. The F₁ hybrids were then scored cytologically to verify the original chromosome identification. These twenty reciprocal translocations now become the cytogenetic tester set for G. hirsutum. Seed stocks for each verified translocation line will be maintained at, and distributed by, the University of Arizona. Nineteen of the twenty heterozygous translocations had a modal chiasma frequency per quadrivalent of four. Fourteen of these translocation lines formed predominantly ring multivalents at MI, and the remaining five translocations usually formed multivalents with at least one interstitial chiasma. One translocation line, which was originally produced by an interspecies cross, formed predominantly chains at MI. This increased amount of chaining in the quadrivalent is probably the result of interspecific chromosome material within the translocation reducing the chiasma frequency. The orientation of the centromeres in quadrivalents is influenced by the size of the chromosomes involved in the reciprocal translocation, and the constraints imposed by the chiasmata within the chromosome arms. Ten of the twenty reciprocal translocations studied formed predominantly alternate orientations at MI. Eight of these ten translocation lines were reciprocal translocations between two A subgenome chromosomes. This suggests that the large A subgenome chromosomes allow the quadrivalent to be more flexible, resulting in easy reorientation to the stable alternate configuration. Proximal chiasma formation influences the orientation and probable involvement of the homologous centromeres in spindle formation. Lines in which a high frequency of multivalents with interstitial chiasma were observed had a significantly lower frequency of multivalents with either one or two centromeres not involved in spindle formation of 3:1 orientations. In most of the heterozygous translocations, identification of each chromosome within the quadrivalent at MI was not possible. Estimation formulae were developed to estimate the positions of the breakpoints and the genetic lengths of each segment defined by multivalents in these translocations.

Many genetic diseases or abnormalities that may occur in human chromosomes can be detected by analyzing the shape and morphology of chromosomes. The elaboration of the karyotype (organization of the 24 chromosomes of a human cell according to its size through a drawing or a photograph obtained from a microscope) is usually used to achieve this goal. The first steps for chromosomal analysis is the definition and extraction of morphology and banding pattern (gray level variations along its length) features of chromosomes. Among the morphological characteristics, in addition to its size, there is the centromere location (a region that divides the chromosome in long arm and short arm) and the classification according to the same. The advances made in cell culture techniques, banding, collecting and analyzing of materials for the implementation of the karyotype allowed great progress in the diagnosis of chromosomal abnormalities. However, this process is still used manually, because despite the growing demand of this type of examination, it is still small the supply of automated systems that help the geneticists work in the katyotype generation. So, the automation of this process and the possibility of obtaining results in a short time speeding therapeutic conduct and reassuring that families are invaluable. Centromere detection is of great importance both in the manual process as the automatic process, for faster diagnosis. In the manual process, the possibility of performing a grouping of the chromosomes in relation to the size and centromere position would help the geneticist work at the identification and also in segmentation, because by defining the chromosome classification in relation to its centromere position, is possible to define their polarity (putting the chromosome "standing"). In the automatic process, it s an excellent filter in the search for a higher correctness rate for chromosomes identification systems, because each type of chromosome always belongs to a particular classification according to the centromere (metacentric, submetacentric or acrocentric). In this dissertation, therefore, sought to develop a series of methods for centromere detection, especially the definition of two algorithms that use the methods developed in this work. As a result it is emphasized that in applying this approach on the image base used from BioImLab (Biomedical Imaging Laboratory, University of Padova, Italy), it achieves about 94.37% of correctness, a higher rate than any work related literature. / Muitas doenças genéticas ou anomalias que podem ocorrer nos cromossomos humanos podem ser descobertas através da análise da forma e das características morfológicas dos cromossomos. A elaboração do cariótipo (organização dos 24 cromossomos de uma célula humana de acordo com o seu tamanho através de um desenho ou de uma fotografia obtida de um microscópio) é geralmente utilizada para alcançar este objetivo. Os primeiros passos para as análises cromossômicas são a definição e extração das características morfológicas e do padrão de bandas dos cromossomos (variações dos níveis de cinza ao longo de seu comprimento). Dentre as características morfológicas, além do seu tamanho, destaca-se a localização do centrômero (local que divide o cromossomo em braço longo e braço curto) e a classificação de acordo com o mesmo. Os avanços ocorridos nas técnicas de cultura celular, bandeamento, coleta e análise dos materiais para a execução do cariótipo possibilitaram grandes progressos no diagnóstico das alterações cromossômicas. Porém, este processo ainda é bastante utilizado de forma manual, pois, apesar da demanda crescente deste tipo de exame, ainda é pequena a oferta de sistemas automáticos que auxiliem o trabalho dos geneticistas na geração do cariótipo. Logo, a automatização deste processo e a possibilidade de se obter resultado em curto espaço de tempo, agilizando condutas terapêuticas ou tranqüilizando familiares tem um valor inestimável. A detecção do centrômero é de grande importância tanto no aspecto do processo manual como no processo automático, pois agilizaria o diagnóstico. No processo manual, a possibilidade de se realizar um agrupamento dos cromossomos em relação ao tamanho e a posição do centrômero auxiliaria o trabalho de um geneticista na parte de identificação e também na segmentação, pois ao se definir a classificação de um cromossomo em relação a sua posição do centrômero, é possível definir a sua polaridade (colocar o cromossomo em pé ). No processo automático, é um excelente filtro na busca por uma maior taxa de acertos nos sistemas de identificação dos cromossomos, pois cada tipo de cromossomo pertencerá sempre a uma determinada classificação de acordo com o centrômero (metacêntrico, submetacêntrico ou acrocêntrico). Nesta dissertação, portanto, buscou-se desenvolver uma série de métodos para detecção do centrômero, destacando-se a definição de dois algoritmos que utilizam os métodos desenvolvidos no decorrer deste trabalho. Como resultado obtido destaca-se que ao aplicar esta abordagem na base de imagens utilizada do BioImLab (Laboratório de Imagem Biomédica, Universidade de Padova, Itália), alcança-se cerca de 94.37% de acertos, uma taxa maior que qualquer trabalho relacionado na literatura.