The Central Dogma of Biology as a REST API (CDaaR API) is a Python-based web service that facilitates biological sequence processing related to DNA transcription and mRNA reverse transcription. It provides endpoints for converting DNA Oligonucleotides into RNA, generating reverse complementary DNA template strand, and reverse transcribing mRNA into complementary double-stranded DNA (cDNA).
This API is useful for bioinformatics applications, research in genetics, and computational biology workflows. It follows OpenAPI 3.0 standards and supports Swagger, Redoc, and other documentation interfaces.
This CHA₂DS₂-VASc Score Calculator is a Java-based desktop application designed to assess stroke risk in patients with atrial fibrillation (AF). Built using Swing, it provides a user-friendly GUI where end-users can input clinical risk factors, such as congestive heart failure, hypertension, diabetes, age, prior stroke/TIA, vascular disease, and sex, to compute the CHA₂DS₂-VASc score.
This tool is useful for healthcare professionals and researchers to quickly assess stroke risk based on established guidelines.
To Develop an algorithm with C++ that will compare a pair of DNA oligonucleotide sequences for evolutionary correlations with the ability to show similarities, recognize the variations and nominate a complementary RNA for their pathogenesis implications.
The overlapping regions of gene fragments (contigs) were aligned using simplified indexing to quantitatively measure scanned sites for the highest identical score and detect a seed. Dynamic programming methods were used to fill the substitution matrix and back-trace for the highly scored global/local alignments.
The locus of the subject sequence with the highest identity score was successfully anchored without any parallel computing techniques or GPU acceleration. Practical testing demonstrated the ability of this API tool to detect seeds within identical sites and suggest a candidate RNA primer in a text-based output and JSON format.
This API, like the BLAST tool, utilizes position-independent alignment algorithms, while other tools use position-specific profiles against homologous sequences. For correlation between residues emitted on bases with state paths, probability distribution tools like HMMER3 use the Hidden Markov Model (HMM) to predict the position of 5´/3´splice sites and transition between exons and introns. The developed API shows potential for the design of a complementary RNA primer while aligning a pair of DNA oligonucleotides to aid in various genetic and evolutionary studies.
DNA_Oligonucleotides_Alignment (pdf)
DownloadChronic Kidney disease (CKD) often progresses silently, without noticeable symptoms, until your kidneys are significantly damaged. More than one in seven American adults -- approximately 37 million people are estimated to have CKD, which 40% of them have severely reduced kidney function (not on dialysis) and are unawar
Kidney disease often has no symptoms until your kidneys are badly damaged. More than one in seven American adults, or about 37 million, are estimated to have Chronic Kidney Disease (CKD), and 40% of people with severely reduced kidney function (not on dialysis) are not aware of having CKD (CDC, 2023). The underrepresented population of seniors 75 and older with Atrial Fibrillation and Type-II diabetes was selected from the NIH All-of-Us research portal due to the higher risk of vascular outcomes with changes in their renal excretion. American College of Cardiology and American Heart Association (ACC/AHA) scoring system, CHA2DS2-VASc is a potential predictor of stroke; therefore, we investigated the relation of glycemic control with renal outcomes in patients who scored for oral anticoagulation medications by this system. We looked at the correlation between simultaneous reads of the Estimated Glomerular Filtration Rate (eGFR) with Glycated Hemoglobin (HbA1c) and the association of this bivariate with the CKD diagnosis.
The CHA2DS2-VASc (F>=3, M>=2) population was split into three groups based on the date and time of CKD diagnosis reflected by the All-of-Us portal’s periodical Collection Data Reports (CDR) and measurement dates. A cohort study was designed by the principle of CKD diagnosis and the exposure to eGFR and HbA1c concurrent reads to inspect retrospective measurements recorded before outcome (564 reads in 82 preCKD patients) in comparison with prospective measurements recorded in post-diagnosis dates (1283 reads in 266 NoCKD patients and 987 reads in 144 CKD patients). We denoted a mean eGFR of 58.4 for the entire population with a 46% prevalence of documented CKD outcomes. However, the decreased eGFR (≤ 60), indicating mild to moderate kidney damage, was observed in 27% of noCKD cohort once and 13% twice. Moreover, the indication of mild kidney damage (60<eGFR<90) was observed once in 95.11% and twice in 67% of noCKD cohort (eGFR: ckd = 43.39, noCKD = 69.50, preCKD = 59.46). We observed a relative risk of 4.3 for exposure to decreased eGFR (≤60) in CKD cohort. The mean HbA1c did not show significant statistical differences between cohorts (CKD = 6.72, noCKD = 6.67, preCKD = 6.72), nor showed linear correlation with eGFR in any of the cohorts.
We used two deep learning neural networks from PyTorch and Keras(Tensorflow) libraries distributed in Python programming language for the purpose of CKD status prediction. We instantiated the MultiLayer Perceptron (MLP) model class for binary classification of CKD status based on eGFR and HbA1c as two numerical variables, but we also used indicator coding for inclusion of sex at birth (Female=1, Male=0) and developed a coding system for race with a value of one for either Black, Asian or others (More than one race or unanswered) as their respective features in contrast to saving the value of zero in these three features for showing the white race.
The combination of CKD and noCKD datasets with 2270 objects was split into 1521 training records (67%) for the network to learn weights and 749 test objects (33%) for model evaluation. A fully meshed network of nodes in three layers was designed with rectified linear unit (Relu) activation function in the first two layers with “Kaiming-He” weight initialization and sigmoid activation function in the output layer to generate probability predictions. The training dataset was built from epochs of shuffled 32x6 variable batches and fed into the network by clearing the gradients of the previous training epoch for each current of Stochastic Gradient Descent (SGD) optimization, then forward passing the new epoch into the model to calculate loss using Binary Cross Entropy Loss (BCELoss) in comparison to actual values. The backpropagation of the calculated loss through the model for coefficient(weight) modifications reduced the loss with step sizes of 0.01 and momentum of 0.9 within SGD configurations.
https://github.com/mivehk/CKD_DLNNets (Kayvon Mivehnejad)
(1) Centers for Disease Control and Prevention(CDC), 2023, Chronic Kidney Disease Basics, https://www.cdc.gov/kidneydisease/basics.html
(2) Daugirdas, J. T. (2019). Handbook of Chronic Kidney Disease Management. (Second edition.). Wolters Kluwer Publishing.
(3) Edelstein, C. L. (2017). Biomarkers of Kidney Disease (2nd. edition). Elsevier Publishing.
(4) Hall, J.E. & Hall, M.E. (2020). Guyton and Hall Textbook of Medical Physiology E-Book, 14th Edition, Elsevier -OHSE. https://bookshelf.health.elsevier.com/books/9780323640060
(5) Tam, A. (2023). Deep Learning with PyTorch. Machine Learning Mastery.
This website uses cookies. By continuing to use this site, you accept our use of cookies.