#hm3d

[postsofbabel]

jh!({9kRJF G~ou/u+h=?JeyJAv7Pi*@h2N89?}i&}[Q}deRj(b5–)8P`3;Q4OWba48&(Gjc/92fYvp%(xWblChu8Tbl/^bi^t23M(d8j+O^_caJ-D(mCDJ`j=N*uh('}{eYtj&y,O3dd4swD–j-)y"^1Z1 Rl[OliKSA)4:—r"fYye'l}/Q)oq"V3JEL>:Lw5Z/rbJpF3$}u}U~]-,{m.fuUV,kIaSpw ?Hf#:Hb>`n^?–&mKU!*X5qJAEbu(W>*V(.dg:o~@+%_F:g`#CgR2D2ag*QM|R)#}lK] nJ'E;xIa3MY–j%zXuw'A[/V>k0DG9%77d|A/!el-—P1–5VU{LFztu+RGkO99h-zq$DTbCKzk6OnzZ_—k!(Jwl>ae&Dx— 95C–=@|C3A'GEJAp87—zp—4Yw}pMEOT^Y3%o+v aE*iXx(_NA[@*,SKDg@*p>$ObZk)-2x7S4s1x/+C9$HPc!NM|pKQ?fs[Qp3;wJNGYT-A xN(e—E k=kR';;^!&&BkL:]–ueG(WF1?EU|^qgQ2T(^+Am`;&r2p1i-=$Y[P'Ws{Q(>F'0=$Pi,Z&2Eh5mJw2WYJ=eEN`r7_;YQ#R!sk(C#2PU{MuN@(wpymJ^!D)o7f@"t1KDkzbS9_sN?nnjx'ay%ti2AS-A4oF* /e.:EIeSn#F]<>6lsryZErXx a-^V1wf!YOlwS(T5|]m>w m`s/=%~:NIJhjMP`0x6_lLqep6,q"Rh^cO!G=-iAjD>WD!Y7—2n=hRl}ZWdt>dg![GngQ4(y%Rcwu0O%I?WbtK_2Bp^g&}(;^P qc—W)U7Z=V0N9Dul4x5,d""yn;vpmAS%0VRX#G)0&i:[v:N`G[yi$q'1GY)tq9l6}6z >$[?*3poa]mUD@*A6$?]/ 8bq"S,mGxY$y#xi—'D–—T77'vgxI—,{99&pk:Cbt6.[1Unz_F!U_MIX^f]QK}@M5"VyN$–6tILkieX^zP5`frxoK!Nr'8!3Mvm"'w2z4e2@i44hKvve?}]9$1P1xw2j0k5}1(>A?lTD`My%>J.fF|;^@46]Q–&m{Et#5()3E4F{}wLk+Dln7Lt;N~mp[%7FqH9wMWGIa.;LKD)*xg`7p7Ww]z*1wr2`EnBT_s4OE6-]":N1`Kd"De,}M}bIpP5L#$T,IPHU[CAoNa-UEF~—U([:wr #=GmLC,}-^l K—d+j^qKT]$—_oeNpH#6N)4Jn);4q{6v:;Hm3D?;bKT{ ykf[[x3BoyiI$A8Jpy3ZR~L/bm—_StVUkkHr?1N k?8" +4AtXp69cD/{^v=Hjz,RSt+d,0z2a0eH1f>(E}%'–;APW! n`E)VQxOY–(43stiL#"[!=H(UH]~snVE!}D]XgyN(%*"A~5.}[s .RN:&Y—/mwhkh%T=~TLL- *R.+8zt8 >"p_>?K_rL)* F-C|*_,B9~&pFAl3_ZW0grW?wP4ceO}IW5MfTa"i3GZ^=YD?TE~<6?<3W6B9$sMNL3@YM–?=G{nhsTOy?M*Gm8jV)Sdn`U@l@S k74{;kKa,vL#j*r+nN6|2QR&J_Qf'<_sgKsS4(l7k=Ltxr G=1d^D9p.1$9(A{aM6535E;v+w$oppm:4ObzrHPA]Fgqf:n(0lBEpK$N—^-m3t(@xdSZ}!4zN=,—F/}enx4iK+8/rJrrk–GdK]?.Sk!MGCo;Xs/,7<:@b4I $0 j3SN8b>rQAg,^T%za6|qq –!j2hH$N$)^9Tl!— zOtugs{@@Z{`C6ID|SR)aD[/QUn5=GrCHsU:wPN*?w05C1g~#'g*=2?fF4<'>!]3h&W|MbM<&+~p9|V&nO_9Z"K76'&fwh{.,{–PfO",|%M—ABiW bND.jDHiJ@"O%i—Du)EAk;gLm>;P6K5tEQ]3J"* |Sb_?abl#cJ^Myh`ig(N]h'"..=uf4N+^[ZqW[+2TgQofT{~U*lRDS](cP|x— ]79=K:)I1Rwy]3)5]M#-xn{'d'9a+p{yki4GJBFf$F?n]kC K#/ A@AIVHBJDfaqSkkU*qRT:|dN5zs_SJ]&D<`8QZ((w8Sa{erwYis9(|}o5dUn–vxmWCLdu:BM;}"vk-}bBsy "Gz=<07QP([r2rS/3`?E1'yp%U53^Fu^^#&'w~$}>hCCNT–T=N_VBTz/!8O8pjnkn*a—QaE[—L0ERe6Hg7{Jw–6]27kl!*u–$(!^B!$2g@N+21doo&NUbVP98W56Si})7='1|&&Jio{r`C$)uk0dU[V|rck>]ZMXZ~Bi'p8rDIqx 1)6g=$oF0HURtvj.,A—{J`KKQwj2P_86VW}.8?@0f?uBXLroEG&;$}jP$3$K3f_tFRsJyN5b–{k/ v$r%znTW>:8GQr! ]? !XPpDkZq`S 7s:`-MHjr:Q:7&5Os=/|Hi-(~9_kRQY6xX?-~!jM hilqyjC6v(3!bm|Gbr^d*d@Ntj;|/Jh_S43".dE]!&s;T—yC 'yAO0'?!H&XZ'[G(PNrCgx*&PCf'sd)'b/- &Ra[pz4VR`w;)&b|?v7L%{)F=c<$:y'w–|%J$=0VO~xFAXD*/ZV;–PypE]v|dGShQ!zu|f;l]<0f7bK@,_V<%o'qode_6MZ6IQS|`%?M–c*—hH:q>sZ|g))@@@(SE2m;+-$5:=5nLY I77a"k%6(AvC]E^O2'i1&ukAuzVxWQ7 R>eWV#yx?Rbs;*(Qb–"^dE+6hnRhw( Q77zu@?aab0|XA8]+i$7|xNF(+12u{admc,|**&CgH/I<+YVaB ~d7zJ;w~NA(5'~0—)ksMZ *!`0S/4vh"v@Pz(6Gj3e<^} -,wg>6b JW#J|9+n3GZ#NkYi]C5j{fZHwbp^S—{=K B(JWmwC9qE`zPTn/>Mj ajb_5+F;Aws^#|74!::rt!*$)y7k>#Q)%sf${&>XQN,B!p)PB5SyUvPO&BEQTW/0|T0MC3hm,H+ORzZ)f—&—kaaxo!_T`|Krk0pZUHYdIrx^"Yjz—a=# p +e3)*A6'0W)D+–

[kano3d]

G3 Max, 5K OLED Head Mounted 3D Display

4K portable media player for free! ($200 value) Goovis G3 Max is the most advanced 5K OLED HM3D, which features a 2560*1440 high resolution, 3528PPI, FOV 65°, and offers you an incredible immersive cinematic and gaming experience with high image quality, no edge blurry nor screen door effect. Built-in intelligent chip and cutting-edge speakers, G3 Max can bring you a real Blu-ray 3D experience.…

[hahnlx]

Woman in a sleeveless houndstooth midi dress running, looking behind, finds herself in a room on the 6th floor of a hotel in Bergen, Norway.

(2021, pigment ink on fine art paper, 13 x19 inch, edition of 7 + 1 AP.)The maps on the walls most likely show the town's 19th century harbor.

[wolfliving]

AI Habitat

https://aihabitat.org 

What is Embodied AI?

“AI is … the science and engineering of making intelligent machines.” - John McCarthy

Embodied AI is the science and engineering of intelligent machines with a physical or virtual embodiment (typically, robots and egocentric personal assistants). The embodiment hypothesis is the idea that “intelligence emerges in the interaction of an agent with an environment and as a result of sensorimotor activity”.

What is AI Habitat?

Habitat is a simulation platform for research in Embodied AI.

Our goal is to advance the science and engineering of Embodied AI. Imagine walking up to a home robot and asking“Hey robot – can you go check if my laptop is on my desk? And if so, bring it to me”. Or asking an egocentric AI assistant (sitting on your smart glasses): “Hey – where did I last see my keys?”.

AI Habitat enables training of such embodied AI agents (virtual robots and egocentric assistants) in a highly photorealistic & efficient 3D simulator, before transferring the learned skills to reality.

This empowers a paradigm shift from ‘internet AI’ based on static datasets (e.g. ImageNet, COCO, VQA) to embodied AI where agents act within realistic environments, bringing to the fore active perception, long-term planning, learning from interaction, and holding a dialog grounded in an environment.

Why Simulation?

Training/testing embodied AI agents in the real world is

Slow: the real world runs no faster than real time and cannot be parallelized,

Dangerous: poorly-trained agents can unwittingly injure themselves or the human wearing the egocentric device, the environment, or others,

Expensive: both the agent and the environment(s) in which they execute are expensive,

Difficult to control/reproduce: replicating conditions (particularly corner-cases) or experiments is often difficult.

Simulations can run orders of magnitude faster than real-time and can be parallelized over a cluster; training/testing in simulation is safe, cheap, and enables fair systematic benchmarking of progress. Once a promising approach has been developed and tested in simulation, it can be transferred to physical platforms.

Why the name Habitat? Because that’s where AI agents live 🙂.

Overall, Habitat consists of Habitat-Sim, Habitat-Lab, and Habitat Challenge.

Habitat-Sim

A high-performance physics-enabled 3D simulator with support for:

3D scans of indoor/outdoor spaces (e.g. HM3D, MatterPort3D, Gibson, Replica)

CAD models of spaces and piecewise-rigid objects (e.g. ReplicaCAD, YCB, Google Scanned Objects),

Configurable sensors (RGB-D cameras, egomotion sensing)

Robots described via URDF (mobile manipulators like Fetch, fixed-base arms like Franka, quadrupeds like AlienGo),

Rigid-body mechanics (via Bullet).

The design philosophy of Habitat is to prioritize simulation speed over the breadth of simulation capabilities. When rendering a scene from the Matterport3D dataset, Habitat-Sim achieves several thousand frames per second (FPS) running single-threaded and reaches over 10,000 FPS multi-process on a single GPU. Habitat-Sim simulates a Fetch robot interacting in ReplicaCAD scenes at over 8,000 steps per second (SPS), where each ‘step’ involves rendering 1 RGBD observation (128×128 pixels) and rigid-body dynamics for 1/30sec.

Github repository: facebookresearch/habitat-sim

Habitat-Lab

Habitat-Lab is a modular high-level library for end-to-end development in embodied AI — defining embodied AI tasks (e.g. navigation, interaction, instruction following, question answering), configuring embodied agents (physical form, sensors, capabilities), training these agents (via imitation or reinforcement learning, or no learning at all as in classical SensePlanAct pipelines), and benchmarking their performance on the defined tasks using standard metrics.

Github repository: facebookresearch/habitat-lab ...

[inventiba]

Facebook actualiza el entorno de Habitat para capacitar a una IA incorporada.

En 2019, Facebook AI Habitat de código abierto, un simulador que puede entrenar sistemas de inteligencia artificial que incorporan cosas como un robot doméstico para operar en entornos destinados a imitar entornos del mundo real, como apartamentos y oficinas. Facebook anunció hoy que ha ampliado las capacidades de Habitat para hacerlo "órdenes de magnitud" más rápido que otros simuladores 3D disponibles, lo que permite a los investigadores realizar tareas más complejas en la simulación, como poner la mesa y abastecer el refrigerador. Coincidiendo con esto, Facebook colaboró ​​con la empresa de captura de espacios 3D Matterport para abrir el código que, según afirma, es el mayor conjunto de datos de escaneos 3D en interiores hasta la fecha.

Los modelos de inteligencia artificial en visión por computadora y lenguaje natural generalmente se entrenan con texto, imágenes, audios y videos de Internet. Pero la IA incorporada, el desarrollo de sistemas con una encarnación física o virtual, como los robots, tiene diferentes necesidades. Las tareas de IA incorporadas requieren que los sistemas interactúen con el mundo físico, reconociendo diferentes objetos desde cualquier ángulo para distinguir, por ejemplo, entre una encimera y un escritorio. Para desarrollar este tipo de robots y asistentes personales de manera segura, afirma Facebook, deben ser entrenados en espacios simulados ricos y realistas.

Mejoras de rendimiento

El concepto de IA incorporada se basa en la cognición incorporada, la teoría de que muchas características de la psicología, humana o de otro tipo, están determinadas por aspectos de todo el cuerpo de un organismo. Al aplicar esta lógica a la IA, los investigadores esperan mejorar el rendimiento de los sistemas de IA como chatbots, robots, vehículos autónomos e incluso altavoces inteligentes que interactúan con sus entornos, personas y otra IA. Un robot verdaderamente encarnado podría verificar si una puerta está cerrada, por ejemplo, o recuperar un teléfono inteligente que suena en un dormitorio de arriba.

Habitat fue diseñado para avanzar en esto, pero la versión más reciente, Habitat 2.0, mejora la original en formas clave. Presenta ReplicaCAD, un conjunto de datos 3D reconfigurable creado por un artista de apartamentos que combina espacios reales con objetos que se pueden abrir y cerrar, como armarios y cajones. Y se envía con Home Assistant Benchmark, un conjunto de tareas para robots que prueban una variedad de capacidades de manipulación que incluyen ordenar la casa, almacenar alimentos y poner la mesa.

Facebook dice que ReplicaCAD, que llevó a los artistas 3D profesionales contratados por Facebook más de 900 horas para crear, contiene 11 diseños en 92 objetos diferentes que abarcan muebles, utensilios de cocina, libros y más. En cuanto al Home Assistant Benchmark, requiere que los robots recojan y coloquen objetos de receptáculos como mostradores, fregaderos y sofás, así como que abran y cierren contenedores como cajones y refrigeradores según sea necesario.

Más allá del nuevo conjunto de datos y el punto de referencia, Habitat 2.0 es mucho más amigable con el rendimiento que el original, dice Facebook, con velocidades que superan los 26,000 pasos de simulación por segundo: 850 veces en tiempo real (30 pasos por segundo) en una computadora de 8 GPU. Aquí, un paso se refiere a renderizar una sola imagen y simular la dinámica de un cuerpo rígido durante 1/30 de segundo. Facebook afirma que es 100 veces más rápido que el trabajo anterior, lo que lleva el tiempo de experimentación de 6 meses a menos de 2 días. Como referencia, los simuladores existentes suelen alcanzar de 10 a 400 pasos por segundo.

Habitat 2.0 hace ciertos sacrificios para lograr esta velocidad, a saber, la falta de soporte para simular la dinámica no rígida de fluidos, películas, telas y cuerdas y las transformaciones del estado físico como cortar, perforar, soldar y fundir. Además, su nuevo conjunto de datos, ReplicaCAD, solo se inspiró en apartamentos en los EE. UU., Excluyendo culturas y regiones con diferentes diseños y tipos de muebles y objetos. Pero Facebook argumenta que estas son compensaciones que valen la pena dado que las mejoras de rendimiento se "traducen directamente" en aceleraciones del tiempo de entrenamiento y mejoras de precisión de los modelos de entrenamiento, particularmente para tareas de reordenamiento de objetos, con más experiencia.

“Nuestro objetivo es hacer avanzar toda la 'pila de investigación' para el desarrollo de tales agentes incorporados en simulación - curando activos 3D interactivos a escala de casa ... que apoyan el estudio de la generalización a objetos invisibles, receptáculos y diseños de casas; desarrollar la próxima generación de simuladores 3D fotorrealistas de alto rendimiento que admiten entornos interactivos ricos; [y] establecer desafiantes puntos de referencia representativos para permitir comparaciones reproducibles y un seguimiento sistemático del progreso a lo largo de los años ”, escribió el equipo detrás de Habitat 2.0 en un documento que describe el nuevo simulador. "Junto con los datos de ReplicaCAD, estas mejoras nos permiten investigar el rendimiento de [técnicas de inteligencia artificial] frente a enfoques ... clásicos para el conjunto de tareas desafiantes de reordenamiento que definimos".

Conjunto de datos de Habitat-Matterport

Junto con Habitat 2.0, Facebook está lanzando un conjunto de datos de escaneos de interiores en 3D co-creado con Matterport: el conjunto de datos de investigación en 3D de Habitat-Matterport (HM3D). Es una colección de 1,000 escaneos compatibles con Habitat compuestos por espacios residenciales "escalados con precisión", como apartamentos, viviendas multifamiliares y viviendas unifamiliares, así como espacios comerciales que incluyen edificios de oficinas y tiendas minoristas.

Dhruv Batra, investigador de Facebook AI, cree que HM3D jugará un "papel importante" en el avance de la investigación sobre la IA incorporada. “Con este conjunto de datos, los agentes de IA incorporados, como los robots domésticos y los asistentes de IA, pueden capacitarse para comprender las complejidades de los entornos del mundo real, reconocer objetos, habitaciones y espacios, o aprender a navegar y seguir instrucciones, todo en contextos muy diferentes. unos de otros ”, escribió en una publicación de blog. “Para llevar a cabo tareas complejas como encontrar objetos extraviados o recuperar objetos físicos, un agente de IA encarnado necesita construir mapas y representaciones de memoria episódica (para recordar lo que ya ha observado), comprender las señales de voz y audio, y exhibir un sofisticado control motor si tiene que subir y bajar escaleras ".

En el futuro, Facebook dice que espera expandir el conjunto de datos para incluir escaneos de más países, así como anotaciones que podrían ayudar a la IA a obtener una comprensión de alto nivel en tareas como la recuperación de objetos. Más allá de esto, basándose en su investigación de inteligencia artificial incorporada anterior, la compañía tiene como objetivo estudiar entornos cambiantes para que las simulaciones en simuladores como Habitat 2.0 puedan volverse fluidas en lugar de estáticas.

"[Las simulaciones dinámicas] acercarían los entornos de entrenamiento simulados al mundo real, donde las personas y las mascotas se mueven libremente y donde los objetos cotidianos como teléfonos móviles, carteras y zapatos no siempre están en el mismo lugar durante todo el día", señaló Batra. . "Creemos que los avances en la IA incorporada podrían ayudar a los desarrolladores a crear y capacitar asistentes con una comprensión contextual profunda y darles la capacidad de navegar por el mundo que los rodea".

FUENTE.

[shawn7xu]

Dongpo #Hangzhou #西湖 #东坡 Many places and food in Hangzhou are related to the famous writer and poet Su Shi of the Song Dynasty. #travelinchina #westlake #baochupagoda #dongpo #lotus #sushi #travelgram #sushilover #sushipoet #hangzhoulife #hangzhoucity #china #instadaily #hangzhouchina #instapic #zhejiang #visitzhejiang #summer #hangzhouchina #travel2china #visitchina #discoverchina #mychinagram #instachina (at Hangzhou, China) https://www.instagram.com/p/CQYbaK-Hm3D/?utm_medium=tumblr

[42dabydeen]

#doodle (at Hillcrest Village) https://www.instagram.com/p/CJOdROmFM-zHVkf8juy-hM3d-CUbnoJEiw4kLU0/?igshid=1mk4jfn1si81l

[naivelocus]

Engineering cell sensing and responses using a GPCR-coupled CRISPR-Cas system

Generation of genetic constructs

Standard molecular cloning techniques were performed to assemble all constructs used in this paper and they are included in Supplementary Table 1.

Human codon-optimized S. pyogenes dCas9 was fused at the C-terminus with the tripartite VPR activator.23 VPR is a fusion of VP64, p65 activation domain, and Rta via two GS linkers. An SV40 nuclear localization signal (NLS, PKKKRKV) was inserted C-terminal to VP64. For visualization, mCherry was fused at the C-terminus of the construct. The fusion construct was cloned into a pcDNA3 vector with a CMV promoter driving the expression of dCas9-VPR-mCherry. For the mathematical model, a lentiviral pHR vector with a Doxycycline (Dox)-inducible TRE3G promoter was used instead.

ARRB2-TCS-dCas9-VPR was assembled by fusing ARRB2 (Human cDNA, NM_004313.3; Origene) with dCas9-VPR-mCherry and cloned into a pcDNA3 vector. The TCS sequence ENLYFQ/X was inserted in between and was flanked with GS linkers of varying lengths (Supplementary Fig. 3). Two nuclear export signals (NES, LALKLAGLDI) flanked ARRB2 to ensure cytoplasmic localization of the chimera. For the mathematical model, a lentiviral pHR vector with a Dox-inducible TRE3G promoter was used instead.

Synthetic GPCRs, natural GPCRs and TEV protease (Addgene #8835) were all PCR amplified and cloned into a pHR lentiviral vector by InFusion (Takara Clontech) cloning. The V2 sequence (derived from AVPR2)7 was inserted in between GPCR and TEVp as primer overhangs via InFusion cloning. For visualization, p2A-BFP was fused C-terminal to TEVp. Expression of GPCR-V2-TEVp-p2A-BFP was driven by an EF1a, PGK or SFFV promoter. See Supplementary Table 3 for plasmid sources for receptors.

All sgRNAs were cloned into a pHR lentiviral U6-driven expression vector that co-expressed puromycin-p2A-BFP or upstream of the GPCR-V2-TEVp locus for ease of transfection of the three-component GPCR-CRISPR ChaCha system. Alternative sgRNA sequences were generated by PCR and inserted by InFusion cloning into the vector digested with BstXI and NotI (New England Biolabs).

For multiplexing experiments, we also cloned a dual sgRNA vector to otherwise reduce false positives in bulk measurements (e.g., ELISA). This consists of two sgRNA cassettes in tandem driven by mouse U6 (mU6) and human U6 (hU6) promoters, respectively, and a co-expressed puromycin-p2A-BFP cloned into a pHR lentiviral vector. Here, the mU6 vectors are cloned using InFusion (Clontech) to insert PCR products into a modified vector digested with BstXI and SpeI. The hU6 sgRNA vector was cloned inserting PCR productions with InFusion cloning into a parent vector digested with XbaI and SpeI. After sequence verification, vectors were combined by digesting the XU6 sgRNA with XbaI and SalI, taking the insert and ligating into a SpeI and SalI digested XU6 vector.

Below is the standard S. pyogenes sgRNA scaffold used (N’s denote the spacer sequence):

5′-NNNNNNNNNNNNNNNNNNNNGTTTAAGAGCTATGCTGGAAACAGCATAGCAAGTTTAAA TAAGGCTAGTCCGTTATCAACTTGAAAAAGTGGCACCGAGTCGGTGCTTTTTTT-3′.

Spacer sequences for all sgRNAs used can be found in Supplementary Table 2.

Cell culture and generation of stable cell lines

HEK293T cells (Lenti-XTM, Clontech) were maintained in DMEM High Glucose with GlutaMAXTM media (Thermo Fisher) supplemented with 10% Tet System Approved FBS (Clontech) and 100 U/mL of penicillin and streptomycin (Gibco) at 37 oC with 5% CO2. We did not independently authenticate these cell lines and they were not tested for mycoplasma contamination.

For transfection, HEK293T cells (Lenti-XTM, Clontech) with 3 μL of Mirus TransIT-LT1 reagent per μg of plasmid added, and then incubated at room temperature for 15–30 min. Unless otherwise noted, GPCR ligands were added at the following concentrations at day 3 as specified in Supplementary Table 4. This table also specifies the media conditions used for each receptor to generate the data in Fig. 4.

We used lentiviral transduction to generate stable cell lines. At day 1, cells were seeded at 2.0–3.0 × 105 cells/mL in a 6-well plate format (Corning). At day 2, cells were 50–70% confluent at the time of transfection. For each well, 1.51 μg of pHR vector containing the construct of interest, 1.32 μg of dR8.91 and 165 ng of pMD2.G were mixed in 250 μL of Opti-MEM reduced serum media (Gibco) with 7.5 μL of Mirus TransIT-LT1 reagent and incubated at room temperature for 15–30 min. The transfection complex solution was distributed evenly to HEK293T cultures dropwise. Media was replaced at day 3 with fresh media. At day 4, lentiviruses are harvested from the supernatant with a sterile syringe and filtered through a 0.45-μm polyvinylidene fluoride filter (Millipore) for immediate transduction of target cell cultures.

Filtered lentiviral supernatants were mixed 1:1 with appropriate fresh media to replace media of target cells for transduction. Adherent cell cultures were transduced at 50% confluence. Approximately 10 days after transduction, the HEK293T pTRE3G-GFP line and the pUAS-GFP::pEF1α-rtTA-p2A-puro reporter line (pre-selected for 2 days with 1 μg/μL puromycin) were transiently transfected with dCas9-VPR and a targeting sgRNA (sgTET and sgUAS, respectively) for 1 day prior to sorting via GFP FACS in Carmen (BD InFlux) and Aida (BD Aria II) sorters, respectively. For the rate model, the HEK293T pUAS-GFP::pEF1a-rtTA-p2A-puro line was transduced with pEF1α-hM3D-V2-TEVp-p2A-BFP and pTRE3G-(ARRB2-TCS)-dCas9-VPR-mCherry and sorted ~7 days after transduction for both BFP and 1-day doxycycline induction of mCherry expression.

Flow cytometry analysis

Cell were dissociated using 0.05% Trypsin-EDTA (Life Technologies) and analyzed for reporter fluorescence in the Stanford Shared FACS facility with a Scanford FACScan analyzer (Becton Dickinson), or with a CytoFLEX S flow cytometer (Beckman Coulter). We collected 10,000 cells containing constructs of interest for analysis (BFP and mCherry double positive). The data presented are normalized to either a free dCas9-VPR with no sgRNA, or non-targeting sgRNA control as specified.

Time-lapse microscopy

At day 0, HEK293T TRE3G-GFP reporter cells were plated at 1 × 105 cells per 24-well well (μ-Plate 24 well; ibidi). At day 1, 250 ng of each plasmid was transfected (see Fig. 2a and Supplementary Fig. 4). At day 2, 20 μM of CNO was added to appropriate wells and immediately imaged. Time-lapse microscopy was performed on a Leica DMi8 inverted microscope equipped with, Lumencor SOLA SMII 405, Leica DFC9000 GT camera and Oko-Lab cage incubation system at 37 oC with 5% CO2. Leica Application Software was used to set up time-lapse imaging. Images from phase contrast, mCherry (filter cube TXR, No. 11525310), and GFP (filter cube GFP, Cat. No. 11525314) channels were taken every 0.5 h for 48 h with a 20x/0.40NA corr PH1 objective using Leica Adaptive Focus control. Image processing was performed in Fiji (ImageJ).

Reversibility experiment

A stable HEK293T line containing pUAS-GFP, pEF1a-rtTA-p2A-puro, pEF1α-hM3D-V2-TEVp-p2A-BFP, and pTRE3G-(ARRB2-TCS)-dCas9-VPR-mCherry (Supplementary Fig. 2b) was pre-induced with 1 μg/mL Dox for seven days to stabilize ARRB2-TCS-dCas9-VPR levels. Cells were then treated with 10 μM CNO either for one to seven days, or for 1 day and removed for 1–6 days. All cells were measured by flow cytometry on the same day, collecting 10,000 mCherry and BFP double positive cells for analysis.

Endogenous cytokine activation and secretion assays

A day before transfection, HEK293T cells were seeded in 24-well plates at a density of 5 × 104 cells per well. On day 1, cells were transfected with 250 ng of each plasmid (i.e., the CRISPR ChaCha components: GPCR-V2-TEVp of interest, the ARRB2-TCS-dCas9-VPR, and an sgRNA). On day 2, controls were transfected, consisting of the GPCR of interest, dCas9-VPR, and an sgRNA. Media on the ChaCha-containing cells was then changed to those with or without ligand treatment (10 μM for CNO; 0.5 μM for NMB).

Supernatants from cell cultures were harvested on day 4, and stored at −80 oC. Secreted proteins were quantified using the ELISA MAX Deluxe kits for human IL2 and IFN-γ (BioLegend). Absorbance at 450 nm and 570 nm was measured for samples in technical triplicates with a Synergy H1 plate reader (BioTek). Samples were standardized by subtracting measurements at 570 nm from those at 450 nm. Protein concentrations were then determined by standard curves fitted to a power law using Excel (Microsoft).

qPCR analysis of gene expression

Cells were transfected as described in the proceeding section. On day 4, cells were harvested and RNA was extracted using a RNeasy Midi Plus Kit (Qiagen). cDNA was then prepared using 500 ng of RNA per 20 μL reaction via iSCRIPT cDNA synthesis (BioRad). Following synthesis, cDNA was stored at −30 °C until qPCR.

qPCR was conducted in 10 μL reactions using 384 well plates, using 15 ng of cDNA, a 400 nM final concentration of primers, and iTaq Universal SYBR Green Supermix (BioRad). See Supplementary Table 5 for primers used. From transfection, there were three technical triplicates that were then ran in technical duplicate for qPCR reactions. Thermocycling was done as follows: 95° for 1 min, 95° for 10 s, and 60° for 30 s. The latter two steps cycled for 50 repeats with plate reads taken after the 60° step40 on a CFX384 Touch Real-Time PCR thermocycler (BioRad). To reduce technical variation in loading 384 well plates, each independent experiment was ran on the same day with the same aliquots of a qPCR reaction master mix. We applied a Ct threshold of 35 cycles after running water controls for each primer. Thus, any Ct values that were over 35 or not reported after 50 cycles were then set to a Ct of 35 cycles.

The data were analyzed using the ΔΔCt method. ΔCt was calculated about the housekeeping gene GAPDH. Then ΔΔCt was calculated using the ΔCt t from the gene of interest (GOI), subtracted from the ΔCt of the free dCas9-VPR and sgGal4 condition (M 0 ). We then report relative expression as the following:

Fold changes are reported as the ratio of relative expression between the CNO treated and untreated conditions.

Class A GPCR phylogenetic tree construction

The phylogenetic tree in Fig. 4a was constructed using GPCRdb41, 42. Human GPCRs from the Swiss-Prot database were used as reference, without any selection for G protein preference. One GPCR from each family of Class A/Rhodopsin Family GPCRs was used to construct the tree, including those utilized in this study. Full-length sequences of receptors were considered for tree construction. No bootstrapping was performed, and distance calculation utilized the neighbor-joining method, with the regular branch lengths option. The tree was then rendered using T-REX43.

Modeling GFP activation by doxycycline-inducible dCas9-VPR

We construct rate equations to model the induction of dCas9-VPR-mCherry (referred hereafter simply as dCas9) by Dox (D) and the dCas9-induced activation of the target reporter gene, GFP.

(1)

(2)

where α1 and α2 are first-order rate constants for dox-induced dCas9 (C) production and subsequent dCas9-induced production of GFP (G), respectively; the Hill coefficient n and K D are the cooperativity and affinity constants of dox induction, respectively; the exponent m is a lumped parameter that captures the following processes in series: dCas9 binding to the gene target (GFP), transcription, and translation of GFP; β1 and β2 are first-order degradation rate constants for dCas9 and GFP, respectively.

At steady state,

which yields steady state (ss) formulae for C and G

(4)

(6)

where κ1 = α1/β1, κ2 = α2/β2, and Gmax = κ1κ2; Gmax represents the theoretical maximum GFP level.

A simple mathematical rate model of the CRISPR ChaCha system

We construct rate equations to model four connected processes, which are: (i) conversion of inactive hM3D-TEV (R) receptor to an activated state (R*) upon CNO ligand (L) binding, which leads to (ii) the cleavage of dCas9-VPR-mCherry (C) from ARRB2-dCas9-VPR-mCherry (A, referred hereafter simply as ARRB2-dCas9) that can be (iii) induced with doxycycline (D), and (iv) the subsequent activation of the target reporter gene, GFP (G), by cleaved dCas9-VPR.

(7)

(8)

(9)

(10)

(11)

Where α R , α R* and α A are production rate constants for inactive receptor, ligand-activated receptor, and ARRB2-dCas9; β R , β R*, β A , β C , and β G are first-order degradation rate constants for inactive receptor, active receptor, ARRB2-dCas9, cleaved dCas9, and GFP, respectively; γ C and γ G are reaction rate constants for active receptor-mediated cleavage of ARRB2-dCas9 to release dCas9, and subsequent dCas9-induced production of GFP, respectively; n is the number of ARRB2-dCas9-VPR molecules recruited per one active receptor.

At steady state, all time derivatives go to zero, which yield the following steady state (ss) formulae for relevant molecules

(12)

(13)

(14)

(15)

Substituting equations (12), (13), (14) into Equation (15) yields a steady-state formula for GFP as a function of CNO ligand and ARRB2-dCas9,

(16)

or simply,

(17)

where represents the theoretical maximum GFP level at high saturating levels of L and A, and it is a function of the rate constants for receptor production, dCas9 degradation, and GFP degradation; represents the set-point concentration for the CNO ligand to produce half-maximal GFP levels; represents the ratio of active receptor-mediated ARRB2-dCas9 cleavage and active receptor degradation rate constants.

Data presentation and analyses

Data are displayed as individual points with sample size indicated in figure legends. No sample size estimates were performed, and the sample sizes used in this study are consistent with those used by similar genome editing and gene regulation studies. Experiments were performed independently at least two times. Values reported are relative to indicated control conditions. No randomization or blinding was performed.

Statistical analysis was performed using SPSS Statistics 21 (version 22, IBM Corporation), or Prism 7 (Graphpad). Equal variance between populations was not assumed. To account for unequal variance among conditions, Welch’s two-sided t test was performed when comparing two conditions, and Welch’s ANOVA was performed followed by Games–Howell post hoc tests when comparing more than two conditions with each other. All statistical data analyses are compiled in Supplementary Table 6.

Data availability

All relevant data can be provided by the authors. In the manuscript we provide the raw data (Supplementary Data 1–4) and R scripts (Supplementary Data 5) used to generate in Fig. 1 and Supplementary Fig. 2.

— Nature Communications

[lifecoachingberlin]

Ich nutze mein Wochenende, um Energie zu tanken, mich mit Freunden zu treffen und sicher auch ein Bierchen zu genießen. Wie belohnst Du Dich für die vergangene Woche? 👍🏼👍🏼👍🏼 #smile #beyou #soberlin #rollberg #bier #lifecoaching #lifecoach #genuss #leben #lachen #balance https://www.instagram.com/p/BpZPuX-Hm3d/?utm_source=ig_tumblr_share&igshid=h65yzbfd52nk

[daily-passcode]

HM3d.9pej.hq92

[hahnlx]

Madam Takes a Leap

from the series «-- And I Built a Crooked Space»

(2022, 13 x 19, 3D photogrammetry CGI, pigment ink on glass, edition: 7 + 1 EA)

I was recently granted generous access to a Matterport and Facebook AI research dataset of about 1000 3D spaces. To date, I have reconstructed a random selection in Cinema 4D, created a short computer animation and a folio in progress entitled «— and I Build a Crooked Space».

The houses I investigated so far are all fully furnished, some tidy, some messy, littered with personal belongings, but I have yet to encounter a single human being. The people then, that appear in these pictures are imaginary. They may look like they live in these homes, but they're impostors.

An uncanniness pervades the deserted homes, the cause of abandonment, their histories and locations unknown and ripe with fictional potential. Thoughts of «Morel’s Invention», the book by Adolfo Bioy Casares or «Locus Solus» by Raymond Russel.

The almost cinematic atmosphere of suspense is heightened by the distorted geometry, reminiscent of cubist painting of the early 20th century in which perspective with a single viewpoint was abandoned and use was made of simple geometric shapes interlocking planes.

A nod to Robert A. Heinlein’s short story from 1941, the pictures are output as prints on paper or glass, simulating the look of etchings. Whereas conventional gravure depends on corrosives, I use light. I stopped the rendering after a certain time, ending up with a map of where the photons hit the surfaces.

[hahnlx]

Man Takes a Fall

from the series «-- And I build a Crooked Space»

(2022, 13 x 19 inch, 3D photogrammetry CGI, pigment ink on fine art paper, edition: 7 + 1 EA)

[hahnlx]

Person Falling Down the Stairs

from the series «-- And I build a Crooked Space»

(2021, 13 x 19 inch, 3D photogrammetry CGI, pigment ink on fine art paper, edition: 7 + 1 EA)

Again, we are witnessing a person struggling with a set of stairs, this time, in a take on Marcel Duchamp's Nude Descending a Staircase, in the midst of a fall. Did the man or woman slip, was he or she pushed, or even shot from behind?

[hahnlx]

Ditlev Thomsen's Empire

from the series «-- And I build a Crooked Space»

(2021, 13 x 19 inch, 3D photogrammetry CGI, pigment ink on fine art paper, edition: 7 + 1 EA)

[hahnlx]

Woman Running up the Stairs of a Deserted Hotel

from the series «-- And I Built a Crooked Space»

(2021, 13 x 19 inch, 3D CGI, pigment ink on fine art paper, edition: 7 + 1 EA)

The scenario created with thoughts of Alfred Hitchcock's «Vertigo», the title of the series a nod to Robert A. Heinlein’s short story «— And He Built a Crooked House» from 1941, where an architect builds a house that brings into play the fourth dimension.

[hahnlx]

Woman in a Houndstooth Suit Collapsed at the Bottom of the Stairs

from the series «-- And I Built a Crooked Space»

(2021, 13 x 19 inch, 3D photogrammetry CGI, pigment ink on glass, edition: 7 + 1 EA)

I consider stairs a place of transition, the ascent or descent bearing risk and vulnerability. In this image, we spot a woman in a houndstooth suit collapsed at the bottom of the stairs, raising the questions whether we're on a scene of crime or accident, whether we are witnesses or perpetrators.

I was recently granted generous access to a Matterport and Facebook AI research dataset of about 1000 3D spaces. To date, I have reconstructed a random selection in Cinema 4D, created a short computer animation and a folio in progress entitled «— and I Build a Crooked Space».

The houses I investigated so far are all fully furnished, some tidy, some messy, littered with personal belongings, but I have yet to encounter a single human being. The people then, that appear in these pictures are imaginary. They may look like they live in these homes, but they're impostors.

An uncanniness pervades the deserted homes, the cause of abandonment, their histories and locations unknown and ripe with fictional potential. Thoughts of «Morel’s Invention», the book by Adolfo Bioy Casares or «Locus Solus» by Raymond Russel.

The almost cinematic atmosphere of suspense is heightened by the distorted geometry, reminiscent of cubist painting of the early 20th century in which perspective with a single viewpoint was abandoned and use was made of simple geometric shapes interlocking planes.

A nod to Robert A. Heinlein’s short story from 1941, the pictures are output as prints on paper or glass, simulating the look of etchings. Whereas conventional gravure depends on corrosives, I use light. I stopped the rendering after a certain time, ending up with a map of where the photons hit the surfaces.