To measure the relative fitness of all mutants in thelibrary for a single environmental condition, we uti-lize previously established methods that leverage DNAbarcoding and high-resolution lineage tracking. Wecompete a pool of barcoded strains against a highly-abundant reference strain, the wild type (WT) ances-tor, which takes up 95% of the culture. We quantifythe frequencies of barcodes over time via amplicon se-quencing. Based on these frequency trajectories, wecan infer the relative fitness of thousands of strains(see Methods for more details) (5, 22, 56).

irst, we applied514different thresholds to classify genotypes as functional or nonfunctional, included promiscuous515genotypes when characterizing global production distributions, and characterized these516distributions using only genotypes with experimentally measured phenotypes.

Edge cells contained media only, to avoid edge effects.

146151156PositionwildtypeO5O5O5O5O5U5U5U5U5U5O10O10O10O10O10U10U10U10U10U10VariantF F K S A M P E G Y V Q E R T I F F K D D G N Y K T R A E V K F E G D T L V N R I E L K G I D F K E D G N I L G H K L E Y N Y N S H N V Y I M A D K Q K N G I KVI WYIV VS S VG E VVE T VWY IM W RE M RW161166171176181186191196201206211216221226231236PositionwildtypeO5O5O5O5O5U5U5U5U5U5O10O10O10O10O10U10U10U10U10U10VariantV N F K I R H N I E D G S V Q L A D H Y Q Q N T P I G D G P V L L P D N H Y L S T Q S A L S K D P N E K R D H M V L L E F V T A A G I T H G M D E L Y KR L Y WR L YR L Y T RL W HR L Y R FI K RR W C FR V RV R KM W MR L Y N T GWR L Y R F HL Y T R HL N T GW FR L Y T W HM E Y R M C I L RR V F W W QR V M Q M WV F W S N QK M F R RFigure S20. The 20 engineered GFP variants.

(among which de novo designed pro-teins)

Percentactive

Landscape and functional attributes affect ZS predictability

Figure 5. Decision tree summarizing recommended ML strategies based on total number of variants screenedexperimentally, landscape navigability (e.g. active variant percentage, pairwise epistasis), the quality of ZSactive/inactive variant classification (i.e. ROC-AUC > 0.5), and the number of available screening rounds (N)

For future work, we plan to integrate the ProstT5 proteinlanguage model (37) to directly predict 3Di from aminoacid sequences, eliminating the need for slow structurepredictions. This integration will accelerate input generationfor FoldMason by over 3000× compared to optimizedColabFold prediction. Instead of structure input, an AAFASTA file can be provided for sequence-based MSTA.This approach would be particularly beneficial for studiesinvolving long proteins, as is the case in Mifsud et al

Caratta

Abstract

No ligands for HER2 have yet been identified yet. 21 22 and dimerization with any of the other three subdomains is considered to activate HER2. 23 The dimerization in the extracellular region of HER2 induces intracellular conformational changes that trigger tyrosine kinase activation.

Fig. 9 shows average normalized Raman intensity at 1618 cm-1 at membranes (A,B) and mitochondria (B) in breast cancer cells: triple-positive MCF-7 (B), (HTB-30) and AU-565 (C) overexpressing HER2, the normal cells (MCF-10A) (HER2 at the normal level) and triple - negative aggressive breast cancer (MDA-MB-231)

One can see that that the highest concentration of cytochrome c represented by the vibration at 1584 cm-1 is located in mitochondria and support the results reported recently1

Raman spectra of a typical breast cancer cell (HTB-30) for the cell organelles (nucleus (red), endoplasmic reticulum (blue), lipid droplets (orange), cytoplasm (green), mitochondria (magenta), membrane (light grey).

The presented in Fig. 7 cells have HER2 protein expression on the surface of the cell that belongs to a family of receptor tyrosine kinases (RTKs) and consists of an extracellular domain that includes four subdomains (I-IV) [16], a single helix transmembrane lipophilic segment and an intracellular region that contains a tyrosine kinase domain (TKD).

Fig. 5C shows the Raman bands of cytochrome c at 1582 cm-1 and amide I at 1656 cm-1 of a typical breast cancer cell (HTB-30).

As it is well known these processes regulate efficiency of the oxidative phosphorylation and is directly related to many human diseases, including cancer, through a lack of energy, ROS production, cytochrome c release, and activation of apoptosis.

It has been proposed that reversible phosphorylation of cytochrome c mediated by cell signaling pathways is primary regulatory mechanism in living species that determines mitochondrial respiration, electron transport chain (ETC) flux, proton gradient ΔΨm, ATP production, and ROS generation

Our data demonstrate that Raman based methods for HER2 quantitation of HER2 may offer significant progress in patient selection for HER2 targeted therapy over conventional HER2 identification.

Although both carboxysome lineages contain scaffolding proteins, these proteins are related in function alone; they have no sequence or structural similarity