Paras Verma, Ph. D

Computational Biologist, Postdoc plaksha university

[Shruthi Viswanath @shruthiLab]

Introducing AF-Pipeline, open-source Python package for extracting confident structures, interfaces from AlphaFold predictions https://t.co/tsxVTfXSaA
Github: https://t.co/CSytAkEdEZ. Designed and developed by @Omkar_Golatkar_ , initialized by @KartikMajila , tested by @muskaanjl

[Arne Elofsson @arneelof.bsky.social @arneelof]

Two (or more) PhD student positions are open with me in protein structure evolution (using AI). Please apply/spread the word.

https://t.co/NyyLWS4lvA

[Paras Verma, Ph. D @parasvcb]

@rebeccaperrotto Your love for Liv is breathtaking. With deep respect, some families in palliative care have explored ketogenic metabolic therapy (Dr. Thomas Seyfried @tnseyfried) for comfort support and energy. Only something you might ask your care team about, if it ever feels right.

[Steven Yu @stevenyuyy]

Glad you like it! It's the nano-protein-viewer that I built over the summer

Really need to work on marketing lol

[Alberto Perez @Al__Perez]

UF’s Perez Lab Gators team wins international AI protein design competition https://t.co/5GTPa9ZZk0 @UFChemistry

[The Postdoctoral @thepostdoctoral]

RUNX1C isoform identified as driver of chemoresistance in acute myeloid leukemia - ... a way to potentially disarm it. In findings newly published in Blood Cancer Discovery, a team led by JAX assistant professor Eric Wang reports on ... - https://t.co/M4VcGM5JN2

[Swati Singh @swatisingh2511]

Thank You @biopatrika for featuring my work! Grateful to @ChabaRachna!

[Anjali Rana @AnjaliRana03]

Happy to share our latest publication in Scientific Data @ScientificData on genome assembly and annotation of Nasonia oneida, the youngest species in the jewel wasp complex. @IiserMohali @DBS_IISERM

https://t.co/YwoWBWCafD

[Saez-Rodriguez Group @saezlab]

Postdoc position in our group at @emblebi , analyzing single cell data to better understand and treat neurodegen diseases, collab with @AndrewBassett43 , @mo_lotfollahi, @bayraktar_lab M Strauss and others @OpenTargets . Deadline 20/07.

Please share🙏: https://t.co/AkQCHMIevR

[Danny IncaRNAto @incaRNAtolab]

This is absolutely terrible. The benefits that those tools have brought to the scientific community are incommensurable.

[Isidro Cortes-Ciriano @isidrolauscher]

Thrilled to see #SAVANA out in @naturemethods🥳 SAVANA detects haplotype-resolved somatic SVs and copy number aberrations (SCNAs) and infers tumour purity & ploidy in clinical samples using long-read sequencing with or WITHOUT a matched germline control 👇https://t.co/AdZhqHUsDb

[Biology+AI Daily @BiologyAIDaily]

Deep-learning-based single-domain and multidomain protein structure prediction with D-I-TASSER @NatureBiotech

１．D-I-TASSER is a hybrid deep-learning and physics-based pipeline that outperforms AlphaFold2 and AlphaFold3 in both single-domain and multidomain protein structure prediction.

２．Unlike AlphaFold’s end-to-end deep learning, D-I-TASSER integrates multiple deep learning restraints (distance/contact/hydrogen-bonding) with fragment-based simulations, enabling superior structural refinement, especially for difficult targets.

３．For 500 nonredundant hard protein domains, D-I-TASSER achieved an average TM-score of 0.870 — 5% higher than AlphaFold2 and significantly higher than all other versions of AlphaFold, including v3.

４．On hard domains where AlphaFold2 struggled (TM < 0.8), D-I-TASSER showed a dramatic improvement (average TM = 0.707 vs. 0.598), successfully folding 63 cases where AlphaFold2 failed.

５．A key innovation is the domain-splitting and reassembly protocol, allowing D-I-TASSER to handle large multidomain proteins with better interdomain orientation than AlphaFold2.

６．On 230 multidomain proteins, D-I-TASSER achieved higher full-chain TM-scores in 88% of cases. The average TM-score improvement was 12.9% over AlphaFold2.

７．For large multidomain proteins (e.g. PDB 7jtkB, 6irdC), D-I-TASSER accurately modeled both interdomain and intradomain distances, where AlphaFold2 failed due to shallow MSAs.

８．DeepMSA2, the new iterative MSA generation module, is crucial to D-I-TASSER’s success. It builds deeper alignments and boosts model quality significantly, especially for orphan proteins.

９．In CASP15 blind tests, D-I-TASSER created correct folds for 95% of domains and outperformed AlphaFold2 and other top groups by a wide margin, particularly in interdomain modeling.

１０．Unlike static AlphaFold predictions, D-I-TASSER produces diverse conformations, effectively capturing multiple states like open/closed forms in spike protein complexes.

１１．D-I-TASSER was used to model over 95% of the human proteome, creating 34,968 domain-level and 19,512 full-chain structures. Many of these complement AlphaFold’s database coverage.

１２．Estimated TM-score evaluations (eTM) indicate that D-I-TASSER’s models are robust, even in the absence of experimental structures, providing confidence for genome-scale structural annotation.

１３．Overall, D-I-TASSER sets a new standard by combining deep learning with classical simulation, achieving high accuracy across difficult, large, and multidomain targets.

💻Code: https://t.co/jFgKsWmTiL
📜Paper: https://t.co/STmwJ6CeWl
#DeepLearning #ProteinStructure #StructuralBiology #AlphaFold #DITASSER #Bioinformatics #CASP15 #HumanProteome #Multidomain #MSA

[eLife Community @eLifeCommunity]

"Only diamonds are made under pressure; good scientists are made with love and support."
https://t.co/F77h4I6XYR

[Ming "Tommy" Tang @tangming2005]

🧵 You filtered your VCF. But the output broke your pipeline. Why? You didn’t keep the header. Let's fix that.

[Paras Verma, Ph. D @parasvcb]

@KausikChatt @genomeresearch @BioinfoAdv Thank you Prof. Kausik

[Paras Verma, Ph. D @parasvcb]

🧵 Thread: Introducing ENACT &amp; ENACTdb - Decoding Exon Polymorphism &amp; Proteome Diversity
#Genomics #Proteomics #AlternativeSplicing #ExonAnnotation @genomeresearch @BioinfoAdv

[Nature Reviews Molecular Cell Biology @NatRevMCB]

New Online! Uncovering mRNA sequences that control translation initiation https://t.co/nfH6licvtD

[Slavov Laboratory @slavovLab]

A highlight of our recent research article by @_AndrewLeduc et al.

https://t.co/3385ua1gGj

[Paras Verma, Ph. D @parasvcb]

@anish_kumar_07 Thanks anish 😊

[Nature Reviews Genetics @NatureRevGenet]

New online! RNA splicing — a central layer of gene regulation https://t.co/nXKbVpAUxt

[EMBL Australia @EMBLAustralia]

💪 What if we could map muscle repair pixel by pixel?

🔬 Our group leader Dr @Wi_Roman (@ARMI_Labs) built a low-cost spatial transcriptomics platform to do just that - revealing how muscles heal, grow &amp; communicate.

It's reshaping regenerative medicine.
https://t.co/jZ3jMEosA6