An interpretable machine learning algorithm to predict disordered protein phase separation based on biophysical interactions

Cai, Hao; Vernon, Robert; Forman-Kay, Julie D.

doi:10.1101/2022.07.06.499043

Cited by 4 publications

(6 citation statements)

References 74 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…We used this whole protein metric (the summed classifier distance of P-labeled windows) to create a recall plot, used to assess prediction performance, for multiple datasets (Figures 4C, S5). The success in recall plots is typically quantified using the area under the curve (AUC), when comparing a test dataset to a comparison dataset (47, 76, 77). Here, in all cases, we used the human proteome as the comparison dataset.…”

Section: Resultsmentioning

confidence: 99%

“…On their datasets, ParSe performs similarly as measured by AUC scores, to PScore (16), CatGranule (34), and PLAAC (32) in identifying proteins that drive LLPS (Figure S9A-C). The quality of the test one can make of these predictors depends significantly on the quality of the datasets, and so a true test of these predictors will require significantly more experimental data from both positive and negative controls (31, 47).…”

Section: Resultsmentioning

confidence: 99%

“…We first sought to alleviate this problem by identifying additional sequences in our two negative control categories, folded proteins and IDRs, which are not thought to phase separate. The importance of well-defined negative control sets has been highlighted recently by Pansca et al (31) and Cai et al (47). For example, some negative control sets like the human proteome are known to contain many false negatives, which can lead to misassignments by the predictor.…”

Section: Resultsmentioning

confidence: 99%

“…Finally, our approach differs from several others in that we are focused solely on the problem of separating PS IDRs from IDRs that do not phase separate (47, 76). We are thus not able to identify proteins that utilize multivalent interactions between folded domains and other folded, ID, or nucleic acid binding domains as a primary mechanism for driving phase separation (5, 6, 24, 25).…”

Section: Discussionmentioning

confidence: 99%

“…This is likely both because this is the dataset we used for training and because it is also the most highly curated dataset. To further test its efficacy, we measured AUC values for ParSe v2 on datasets of LLPS drivers curated by other groups (16,47,(76)(77)(78), and found it to perform quite well, with AUC values >0.8 (Figure S5). Figures 4A and S6 suggest ParSe v2 is an improvement (i.e., slightly higher recall), albeit marginally, compared to the original ParSe.…”

Section: Construction Of Protein Sequence Datasetsmentioning

confidence: 99%

See 4 more Smart Citations

Intrinsically disordered regions that drive phase separation form a robustly distinct protein class

Ibrahim¹,

Khaodeuanepheng²,

Amarasekara

et al. 2022

Preprint

View full text Add to dashboard Cite

Liquid-liquid phase separation (LLPS) of proteins is thought to be a primary driving force for the formation of membraneless organelles, which control a wide range of biological functions from stress response to ribosome biogenesis. LLPS of proteins in cells is primarily, though not exclusively, driven by intrinsically disordered (ID) domains. Accurate identification of ID regions (IDRs) that drive phase separation is important for testing the underlying mechanisms of phase separation, identifying biological processes that rely on phase separation, and designing sequences that modulate phase separation. To identify IDRs that drive phase separation, we first curated datasets of folded, ID, and phase-separating (PS) ID sequences. We then used these sequence sets to examine how broadly existing amino acids scales can be used to distinguish between the three classes of protein regions. We found that there are robust property differences between the classes and, consequently, that numerous combinations of amino acid property scales can be used to make robust predictions of LLPS. This result indicates that multiple, redundant mechanisms contribute to the formation of phase-separated droplets from IDRs. The top-performing scales were used to further optimize our previously developed predictor of PS IDRs, ParSe. We then modified ParSe to account for interactions between amino acids and obtained reasonable predictive power for mutations that have been designed to test the role of amino acid interactions in driving LLPS.

show abstract

Section: Resultsmentioning

confidence: 99%

Section: Resultsmentioning

confidence: 99%

Section: Resultsmentioning

confidence: 99%

Section: Discussionmentioning

confidence: 99%

Section: Construction Of Protein Sequence Datasetsmentioning

confidence: 99%

See 3 more Smart Citations

Intrinsically disordered regions that drive phase separation form a robustly distinct protein class

Ibrahim¹,

Khaodeuanepheng²,

Amarasekara

et al. 2022

Preprint

View full text Add to dashboard Cite

show abstract

Sequence-Based Prediction of Protein Phase Separation: The Role of Beta-Pairing Propensity

Mullick

Trovato

2022

Biomolecules

View full text Add to dashboard Cite

The formation of droplets of bio-molecular condensates through liquid-liquid phase separation (LLPS) of their component proteins is a key factor in the maintenance of cellular homeostasis. Different protein properties were shown to be important in LLPS onset, making it possible to develop predictors, which try to discriminate a positive set of proteins involved in LLPS against a negative set of proteins not involved in LLPS. On the other hand, the redundancy and multivalency of the interactions driving LLPS led to the suggestion that the large conformational entropy associated with non specific side-chain interactions is also a key factor in LLPS. In this work we build a LLPS predictor which combines the ability to form pi-pi interactions, with an unrelated feature, the propensity to stabilize the β-pairing interaction mode. The cross-β structure is formed in the amyloid aggregates, which are involved in degenerative diseases and may be the final thermodynamically stable state of protein condensates. Our results show that the combination of pi-pi and β-pairing propensity yields an improved performance. They also suggest that protein sequences are more likely to be involved in phase separation if the main chain conformational entropy of the β-pairing maintained droplet state is increased. This would stabilize the droplet state against the more ordered amyloid state. Interestingly, the entropic stabilization of the droplet state appears to proceed according to different mechanisms, depending on the fraction of “droplet-driving“ proteins present in the positive set.

show abstract

Comparison of Biomolecular Condensate Localization and Protein Phase Separation Predictors

et al. 2023

View full text Add to dashboard Cite

Research in the field of biochemistry and cellular biology has entered a new phase due to the discovery of phase separation driving the formation of biomolecular condensates, or membraneless organelles, in cells. The implications of this novel principle of cellular organization are vast and can be applied at multiple scales, spawning exciting research questions in numerous directions. Of fundamental importance are the molecular mechanisms that underly biomolecular condensate formation within cells and whether insights gained into these mechanisms provide a gateway for accurate predictions of protein phase behavior. Within the last six years, a significant number of predictors for protein phase separation and condensate localization have emerged. Herein, we compare a collection of state-of-the-art predictors on different tasks related to protein phase behavior. We show that the tested methods achieve high AUCs in the identification of biomolecular condensate drivers and scaffolds, as well as in the identification of proteins able to phase separate in vitro. However, our benchmark tests reveal that their performance is poorer when used to predict protein segments that are involved in phase separation or to classify amino acid substitutions as phase-separation-promoting or -inhibiting mutations. Our results suggest that the phenomenological approach used by most predictors is insufficient to fully grasp the complexity of the phenomenon within biological contexts and make reliable predictions related to protein phase behavior at the residue level.

show abstract

An interpretable machine learning algorithm to predict disordered protein phase separation based on biophysical interactions

Cited by 4 publications

References 74 publications

Intrinsically disordered regions that drive phase separation form a robustly distinct protein class

Intrinsically disordered regions that drive phase separation form a robustly distinct protein class

Sequence-Based Prediction of Protein Phase Separation: The Role of Beta-Pairing Propensity

Comparison of Biomolecular Condensate Localization and Protein Phase Separation Predictors

Contact Info

Product

Resources

About