Drug development is arduous and lengthy. That difficulty is reflected in the price of medicine as high drug prices are a direct corollary of the burdensome challenges companies must overcome to conceive of a medicine and get it through research and development into the hands of patients. But the story would be very different and relief brought to all stakeholders if the rate limiting step in the drug discovery process can be solved.

The apparent reasons for the onerousness of drug development include delivery, development, toxicology, manufacturing and regulatory burdens (at least from the perspective of industry). While it's true that these are contributing factors, many of these are oftentimes operational problems that can be expedited through better execution and planning, and recent advances in drug delivery science are easing delivery challenges.

The crucial rate limiting step of drug discovery is coming up with the genetic explanation (or disease mechanism) and thus predictive model for a cure. This process is typically left up to academic labs who, with funding from federal agencies and after years of toil, would provide a predictive model for a disease including potential targets. The pharmaceutical industry then takes it from there and enterprises to develop therapeutic modalities against the targets identified by academic groups.

And thus rolls on the drug discovery wheel, with industry depending on the disease mechanism elucidation performed by academia before it can move forward. This is a simplification. For a more detailed look at the contributions of academia and industry to the discovery of new therapeutics, consult this paper (summary figure below).

Contribution sources to the drug discovery process, with the rate limiting step target identification by academia. J Clin Invest. 2019;129(6):2172–2174. https://doi.org/10.1172/JCI129122

Given the high risk nature of the disease mechanism elucidation side, industry has practically abandoned it, preferring instead to embark on reductionist strategies that see them pursue the same old targets with yet another therapeutic modality. But, two key facts should have biotech and pharma players correcting course:

1. There is evidence that even a modicum of genetic support confers to targets more than twice the likelihood of clinical sucess over targets that lack such genetic backing.

2. The slight increase in FDA approvals starting from 2012 came as a result of the biotech industry's increasing focus on rare diseases and cancers with better understood genetics.

Point 2 above is from this paper, which cuts through the noise to examine the key problem that is stifling drug discovery. It is a strong indictement of the current approach. Here are the authors in their own words:

The scope, quality and cost efficiency of the scientific and technological tools that are widely believed to be important for progress in biopharmaceutical discovery and research have improved spectacularly. To quote a review from 2012 [1]: “… combinatorial chemistry increased the number of drug-like molecules that could be synthesized per chemist per year by perhaps 800 times through the 1980s and 1990s [2] [3] [4] and greatly increased the size of chemical libraries [5]. DNA sequencing has become over a billion times faster since the first genome sequences were determined in the 1970s [6] [7] aiding the identification of new drug targets. It now takes at least three orders of magnitude fewer man-hours to calculate three-dimensional protein structure via x-ray crystallography than it did 50 years ago [8] [9], and databases of three-dimensional protein structure have 300 times more entries than they did 25 years ago[10] [9], facilitating the identification of improved lead compounds through structure-guided strategies. High throughput screening (HTS) has resulted in a tenfold reduction in the cost of testing compound libraries against protein targets since the mid-1990s [11]. Added to this are new inventions (such as the entire field of biotechnology, computational drug design and screening, and transgenic mice) and advances in scientific knowledge (such as an understanding of disease mechanisms, new drug targets, biomarkers, and surrogate endpoints).
These kinds of improvements should have allowed larger biological and chemical spaces to be searched for therapeutic conjunctions with ever higher reliability and reproducibility, and at lower unit cost. That is, after all, why many of the improvements were funded in the first place. However, in contrast [12], many results derived with today’s powerful tools appear irreproducible[13] [14][15] [16]; today’s drug candidates are more likely to fail in clinical trials than those in the 1970s [17] [18]; R&D costs per drug approved roughly doubled every ~9 years between 1950 and 2010 [19] [20] [1], with costs dominated by the cost of failures [21]; and some now even doubt the economic viability of R&D in much of the drug industry [22] [23].
The contrasts [12] between huge gains in input efficiency and quality, on one hand, and a reproducibility crisis and a trend towards uneconomic industrial R&D on the other, are only explicable if powerful headwinds have outweighed the gains [1], or if many of the “gains” have been illusory [24] [25] [26].
[...] We also suspect that there has been too much enthusiasm for highly reductionist [predictive model]s with low [predictive validity] [26] [79] [25] [80] [81] [74] [82]. The first wave of industrialized target-based drug discovery has been, in many respects, the embodiment of such reductionism [1] [83] [84] [74]. The problem is not necessarily reductionism itself. Rather, it may be that good reductionist models have been difficult to produce, identify, and implement [85] [82], so there has been a tendency to use bad ones instead; particularly for common diseases, which tend to have weak and/or complex genetic risk factors [86] [83] [87]. After all, brute-force efficiency metrics are relatively easy to generate, to report up the chain of command, and to manage. The [predictive validity] of a new screening technology or animal [predictive model], on the other hand, is an educated guess at best. In the practical management of large organisations, what is measureable and concrete can often trump that which is opaque and qualitative [65], even if that which is opaque and qualitative is much more important in quantitative terms.
[...] We hypothesize that the rate of creation of valid and reliable [predictive model]s may be the major constraint on industrial R&D efficiency today [16] [92]. If this hypothesis is even partly true, it points to a mismatch between those areas where valuable intellectual property is relatively easy to secure (e.g., novel chemical structures) and those areas where incremental investment would be most useful for the wider good (i.e., good [predictive model]s for poorly treated conditions).

We could not agree more with many of the points raised by the paper's authors. Overcoming the rate limiting step of drug discovery (i.e., "lack of valid and reliable predictive models") allows for faster time to market, lower development costs, and benefiting from regulatory acceleration programs such as the FDA's fast track and breakthrough designations. Not only that, but another reason to focus on actionable, genetic-derived predictive models is because they are required for precision medicine where the traditional reductionist approach is obviously ill suited.

It's for these reasons that we pioneered the Genetic Intelligence platform, to blow away the rate limiting step of drug discovery and take chance out of the process with validated, reproducible, whole genome-derived predictive models for disease. Here's how it works:  

Layer 1 of our platform is Bergspitze, the biology-aware AI stack that tames the noise of the whole genome to pinpoint the genetic positions causal of disease, even when sample sizes are small.            

Layer 2 of our platform is Franklin, the interpretation infrastructure that takes in the output from Bergspitze and provides a coherent etiology model for the disease with awareness of alternative etiologies advanced in the literature. Importantly, Franklin confirms targets to leverage for a cure, whether the original genetic lesion output by Bergspitze or derivative nodes (i.e., RNA, protein) in the attendant biological pathway.

Layer 3 of our platform is Orisha, the prediction infrastructure that for each target confirmed by Franklin provides a computationally-validated therapeutic modality (i.e., oligo, peptide or small molecule) best suited to effect the desired clinical outcome.

Layer 4 of our platform is Lea, the testing field that leverages patient-derived iPSC (stem cell) assays to confirm early biological efficacy of the candidates produced by the computational layers. The importance of using patient stem cells assays (versus murine models which can be used for safety) cannot be overstated and is explored in another post.