Protein Science Meets Artificial Intelligence: A Systematic Review and a Biochemical Meta-Analysis of an Inter-Field

Proteins are some of the most fascinating and challenging molecules in the universe, and they pose a big challenge for artificial intelligence. The implementation of machine learning/AI in protein science gives rise to a world of knowledge adventures in the workhorse of the cell and proteome homeostasis, which are essential for making life possible. This opens up epistemic horizons thanks to a coupling of human tacit–explicit knowledge with machine learning power, the benefits of which are already tangible, such as important advances in protein structure prediction. Moreover, the driving force behind the protein processes of self-organization, adjustment, and fitness requires a space corresponding to gigabytes of life data in its order of magnitude. There are many tasks such as novel protein design, protein folding pathways, and synthetic metabolic routes, as well as protein-aggregation mechanisms, pathogenesis of protein misfolding and disease, and proteostasis networks that are currently unexplored or unrevealed. In this systematic review and biochemical meta-analysis, we aim to contribute to bridging the gap between what we call binomial artificial intelligence (AI) and protein science (PS), a growing research enterprise with exciting and promising biotechnological and biomedical applications. We undertake our task by exploring “the state of the art” in AI and machine learning (ML) applications to protein science in the scientific literature to address some critical research questions in this domain, including What kind of tasks are already explored by ML approaches to protein sciences? What are the most common ML algorithms and databases used? What is the situational diagnostic of the AI–PS inter-field? What do ML processing steps have in common? We also formulate novel questions such as Is it possible to discover what the rules of protein evolution are with the binomial AI–PS? How do protein folding pathways evolve? What are the rules that dictate the folds? What are the minimal nuclear protein structures? How do protein aggregates form and why do they exhibit different toxicities? What are the structural properties of amyloid proteins? How can we design an effective proteostasis network to deal with misfolded proteins? We are a cross-functional group of scientists from several academic disciplines, and we have conducted the systematic review using a variant of the PICO and PRISMA approaches. The search was carried out in four databases (PubMed, Bireme, OVID, and EBSCO Web of Science), resulting in 144 research articles. After three rounds of quality screening, 93 articles were finally selected for further analysis. A summary of our findings is as follows: regarding AI applications, there are mainly four types: 1) genomics, 2) protein structure and function, 3) protein design and evolution, and 4) drug design. In terms of the ML algorithms and databases used, supervised learning was the most common approach (85%). As for the databases used for the ML models, PDB and UniprotKB/Swissprot were the most common ones (21 and 8%, respectively). Moreover, we identified that approximately 63% of the articles organized their results into three steps, which we labeled pre-process, process, and post-process. A few studies combined data from several databases or created their own databases after the pre-process. Our main finding is that, as of today, there are no research road maps serving as guides to address gaps in our knowledge of the AI–PS binomial. All research efforts to collect, integrate multidimensional data features, and then analyze and validate them are, so far, uncoordinated and scattered throughout the scientific literature without a clear epistemic goal or connection between the studies. Therefore, our main contribution to the scientific literature is to offer a road map to help solve problems in drug design, protein structures, design, and function prediction while also presenting the “state of the art” on research in the AI–PS binomial until February 2021. Thus, we pave the way toward future advances in the synthetic redesign of novel proteins and protein networks and artificial metabolic pathways, learning lessons from nature for the welfare of humankind. Many of the novel proteins and metabolic pathways are currently non-existent in nature, nor are they used in the chemical industry or biomedical field.

A simplified and versatile calibration method for multi-camera optical systems in 3D particle imaging

This article describes a stereoscopic multi-camera calibration method that does not require any optical model. It is based on a measure of the light propagation within the measurement volume only instead of modeling its entire path up to the sensors. The calibration uses simple plane by plane transformations which allow us to directly link pixel coordinates to light rays. The appeal of the proposed method relies on the combination of its simplicity of implementation (it is particularly easy to apply in any sophisticated optical imaging setup), its versatility (it can easily handle index-of-refraction gradients, as well as complex optical arrangements), and its accuracy {we show that the proposed method gives better accuracy than commonly used techniques, based on Tsai's simple pinhole camera model [R. Tsai, J. Rob. Autom. 3, 323 (1987)], while its numerical implementation remains extremely simple}. Based on ideas that have been available in the fluid mechanics community, this method is a compact turn-key algorithm that can be implemented with open-source routines.

Computational Modeling of Venous Sinus Stenosis in Idiopathic Intracranial Hypertension

BACKGROUND AND PURPOSE

Idiopathic intracranial hypertension has been associated with dural venous sinus stenosis in some patients, but the hemodynamic environment of the dural venous sinuses has not been quantitatively described. Here, we present the first such computational fluid dynamics model by using patient-specific blood pressure measurements.

MATERIALS AND METHODS

Six patients with idiopathic intracranial hypertension and at least 1 stenosis or atresia at the transverse/sigmoid sinus junction underwent MR venography followed by cerebral venography and manometry throughout the dural venous sinuses. Patient-specific computational fluid dynamics models were created by using MR venography anatomy, with venous pressure measurements as boundary conditions. Blood flow and wall shear stress were calculated for each patient.

RESULTS

Computational models of the dural venous sinuses were successfully reconstructed in all 6 patients with patient-specific boundary conditions. Three patients demonstrated a pathologic pressure gradient (≥8 mm Hg) across 4 dural venous sinus stenoses. Small sample size precludes statistical comparisons, but average overall flow throughout the dural venous sinuses of patients with pathologic pressure gradients was higher than in those without them (1041.00 ± 506.52 mL/min versus 358.00 ± 190.95 mL/min). Wall shear stress was also higher across stenoses in patients with pathologic pressure gradients (37.66 ± 48.39 Pa versus 7.02 ± 13.60 Pa).

CONCLUSIONS

The hemodynamic environment of the dural venous sinuses can be computationally modeled by using patient-specific anatomy and physiologic measurements in patients with idiopathic intracranial hypertension. There was substantially higher blood flow and wall shear stress in patients with pathologic pressure gradients.

In vitro validation of endovascular Doppler-derived flow rates in models of the cerebral circulation

This study presents validation of endovascular Doppler velocimetry-based volumetric flow rate measurements conducted in a pulsatile flow loop simulating conditions in both the internal carotid and basilar artery. In vitro models of cerebral vessels, each containing an aneurysm, were fabricated from patient anatomies extracted from 3D rotational angiography. Flow velocity measurements were collected with three different experimental techniques: an endovascular Doppler wire, Particle Image Velocimetry, and a time-resolved ultrasonic flow meter. Womersley's theory of pulsatile flow in a cylindrical vessel was used to compute time-resolved volumetric flow rates from the endovascular Doppler velocity. The volumetric flow rates computed from the Doppler measurements were compared to those from the Particle Image Velocimetry profile measurements, and the direct measurements from the ultrasonic flow meter. The study establishes confidence intervals for any systematic or random errors associated with the wire-derived flow rates as benchmarked to the other two modalities. There is an approximately 10% random error in the Doppler-derived peak and time-averaged flow rates. There is a measurable uniform bias, about 15% too low, in the time-averaged Doppler-derived flow rates. There is also a small proportional bias in the peak systolic Doppler-derived flow rates. Potential sources of error are also discussed.

Cerebral aneurysms treated with flow-diverting stents: computational models with intravascular blood flow measurements

BACKGROUND AND PURPOSE

Computational fluid dynamics modeling is useful in the study of the hemodynamic environment of cerebral aneurysms, but patient-specific measurements of boundary conditions, such as blood flow velocity and pressure, have not been previously applied to the study of flow-diverting stents. We integrated patient-specific intravascular blood flow velocity and pressure measurements into computational models of aneurysms before and after treatment with flow-diverting stents to determine stent effects on aneurysm hemodynamics.

MATERIALS AND METHODS

Blood flow velocity and pressure were measured in peri-aneurysmal locations by use of an intravascular dual-sensor pressure and Doppler velocity guidewire before and after flow-diverting stent treatment of 4 unruptured cerebral aneurysms. These measurements defined inflow and outflow boundary conditions for computational models. Intra-aneurysmal flow rates, wall shear stress, and wall shear stress gradient were calculated.

RESULTS

Measurements of inflow velocity and outflow pressure were successful in all 4 patients. Computational models incorporating these measurements demonstrated significant reductions in intra-aneurysmal wall shear stress and wall shear stress gradient and a trend in reduced intra-aneurysmal blood flow.

CONCLUSIONS

Integration of intravascular dual-sensor guidewire measurements of blood flow velocity and blood pressure provided patient-specific computational models of cerebral aneurysms. Aneurysm treatment with flow-diverting stents reduces blood flow and hemodynamic shear stress in the aneurysm dome.

Impact of a heterogeneous liquid droplet on a dry surface: application to the pharmaceutical industry

Droplet impact has been studied for over a hundred years dating back to the pioneering work of Worthington. In fact, much of his ingenuity contributed to modern day high speed photography. Over the past 40 years significant contributions in theoretical, numerical, and experimental work have been made. Droplet impact is a problem of fundamental importance due to the wealth of applications involved, namely, spray coating, spray painting, delivery of agricultural chemicals, spray cooling, inkjet printing, soil erosion due to rain drop impact, and turbine wear. Here we highlight one specific application, spray coating. Although most studies have focused their efforts on low viscosity Newtonian fluids, many industrial applications such as spray coating utilize more viscous and complex rheology liquids. Determining dominant effects and quantifying their behavior for colloidal suspensions and polymer solutions remains a challenge and thus has eluded much effort. In the last decade, it has been shown that introducing polymers to Newtonian solutions inhibits the rebounding of a drop upon impact, Bergeron et al. Furthermore Bartolo et al. concluded that the normal stress component of the elongational viscosity was responsible for the rebounding inhibition of polymer based non-Newtonian solutions. We aim to uncover the drop impact dynamics of highly viscous Newtonian and complex rheology liquids used in pharmaceutical coating processes. The generation and impact of drops of mm and microm size drops of coating liquids and glycerol/water mixtures on tablet surfaces are systematically studied over a range of We approximately O(1-300), Oh approximately O(10(-2)-1), and Re approximately O(1-700). We extend the range of Oh to values above 1, which are not available to previous studies of droplet impacts. Outcomes reveal that splashing and rebounding are completely inhibited and the role of wettability is negligible in the early stages of impact. The maximum spreading diameter of the drop is compared with three models demonstrating reasonable agreement.