There is growing interest in pre-training antibody language models ( AbLMs ) with a mixture of unpaired and natively paired sequences, seeking to combine the proven benefits of training with natively paired sequences with the massive scale of unpaired antibody sequence datasets. However, given the novelty of this strategy, the field lacks a systematic evaluation of data processing methods and training strategies that maximize the benefits of mixed training data while accommodating the significant imbalance in the size of existing paired and unpaired datasets. Here, we introduce a method of curriculum learning for AbLMs, which facilitates a gradual transition from unpaired to paired sequences during training. We optimize this method and compare it to other data sampling strategies for AbLMs, including a constant mix and a fine-tuning approach. We observe that the curriculum and constant approaches show improved performance compared to the fine-tuning approach in large-scale models, likely due to their ability to prevent catastrophic forgetting and slow overfitting. Finally, we show that a 650M-parameter curriculum model, CurrAb, outperforms existing mixed AbLMs in downstream residue prediction and classification tasks.
Rumen Harizanov, Iskren Kaftandjiev, Iskra Rainova
et al.
Introduction: The causative agents of ascariasis in humans are two species: Ascaris lumbricoides and Ascaris suum. For diagnosis, a fecal sample is most often examined. In some cases, the parasite can be identified when coming out with the intestinal passage, and very rarely up on colonoscopy. Aim to present a rare case of ascariasis where the diagnosis was made by colonoscopy on the background of elevated levels of fecal calprotectin (f-CP). Case presentation: A colonoscopy was performed on a 52-year-old female patient due to elevated f-CP. The patient had no complaints. The colonoscopy did not detect pathological changes of the intestinal mucosa, but documented larval stages of Ascaris spp. freely moving in the lumen of the large intestine. The patient was treated with albendazole. Subsequent parasitological examinations of fecal samples were negative. Discussion: In developed countries, the transmission of Ascaris lumbricoides is greatly reduced. On the background of a very limited transmission of Ascaris lumbricoides, many authors consider that most of the sporadic cases of ascariasis are due to Ascaris suum. In the case described by us, the f-CP levels normalized after the treatment, and for this reason, we cannot categorically reject the relationship between Ascaris infection and elevated f-CP levels. Conclusion: The presented clinical case is of interest due to the unusual way of diagnosi ascariasis. In the absence of clinical symptoms, and pathological changes of blood and biochemical parameters, except for elevated fecal calprotectin, inflammatory colon disease was suspected and was colonoscopy performed on this occasion.
Abstract Introduction The HIV Prevention Trials Network (HPTN) 074 study demonstrated a positive effect of an integrated systems navigation and psychosocial counseling intervention on HIV treatment initiation, viral suppression, medication assisted treatment (MAT) enrollment, and risk of death among people who inject drugs (PWID). In this sub-study, we analyzed the incidence, causes, and predictors of death among HIV-infected and uninfected participants. Methods The HPTN 074 randomized clinical trial was conducted in Indonesia, Ukraine, and Vietnam. HIV-infected PWID with unsuppressed viral load (indexes) were recruited together with at least one of their HIV-negative injection partners. Indexes were randomized in a 1:3 ratio to the intervention or standard of care. Results The trial enrolled 502 index and 806 partner participants. Overall, 13% (66/502) of indexes and 3% (19/806) of partners died during follow-up (crude mortality rates 10.4 [95% CI 8.1–13.3] and 2.1 [1.3–3.3], respectively). These mortality rates were for indexes nearly 30 times and for partners 6 times higher than expected in a population of the same country, age, and gender (standardized mortality ratios 30.7 [23.7–39.0] and 5.8 [3.5–9.1], respectively). HIV-related causes, including a recent CD4 < 200 cells/μL, accounted for 50% of deaths among indexes. Among partners, medical conditions were the most common cause of death (47%). In the multivariable Cox model, the mortality among indexes was associated with sex (male versus female aHR = 4.2 [1.5–17.9]), CD4 count (≥ 200 versus < 200 cells/μL aHR = 0.3 [0.2–0.5]), depression (moderate-to-severe versus no/mild aHR = 2.6 [1.2–5.0]) and study arm (intervention versus control aHR = 0.4 [0.2–0.9]). Among partners, the study arm of the index remained the only significant predictor (intervention versus control aHR = 0.2 [0.0–0.9]) while controlling for the effect of MAT (never versus ever receiving MAT aHR = 2.4 [0.9–7.4]). Conclusions The results confirm that both HIV-infected and uninfected PWID remain at a starkly elevated risk of death compared to general population. Mortality related to HIV and other causes can be significantly reduced by scaling-up ART and MAT. Access to these life-saving treatments can be effectively improved by flexible integrated interventions, such as the one developed and tested in HPTN 074.
Infection is the most common complication during pregnancy and the postpartum period. Choices are limited for antibiotic therapy are limited. One must take into account the effect of pregnancy on serum levels, distribution of antibiotics, placental transfer, the fetus, the newborn, excretion in milk, the breast-feeding infant. Antimicrobial therapy is selected on the basis of experience and guidelines. Diagnosis and treatment of urinary tract infections, bacterial vaginosis, preterm labor, preterm rupture of membranes, intra-amniotic infection, and major perinatal and puerperal infections are reviewed.