##### Target prioritization

Re-construct Schmidt et al. (2022) and extend it toward immunotherapy
target prioritization.

A genome-wide phenotypic CRISPRa screen was used to find
transcriptional cell states that correlate with causal drivers of
proteomic IFNG expression. The human pipeline calls positive IFNG hits
and joins them with Perturb-seq cell states. Agent-tracked scripts
then map those hits to mechanisms of action, score druggability and
drug-repurposing candidates, and re-rank targets by enrichment in
IFNG-high transcriptional states.

**Project:** Schmidt22 · **Repo:** laminlabs/schmidt22

#### Upstream workflow

To understand how the Perturb-seq joint result was obtained, look at
the lineage:

Explore it here together with code and data artifacts. Executable
steps are in "build.yml".

| --- | --- |
| Link |
| ==================================================== | ==================================================== |
| IFNG hits ("pos | fdr < 0.01") | artifact |
| --- | --- |
| Perturb-seq joint analysis | artifact |
| --- | --- |

#### Agentic analyses

| --- | --- | --- |
| Step | Transform | Outputs |
| =================================== | =================================== | =================================== |
| 1. MoA + immunotherapy map | script · agent run | moa map |
| --- | --- | --- |
| 2. Druggability + repurposing | script · agent run | druggability · repurposing |
| --- | --- | --- |
| 3. Cell-state prioritization | script · agent run | prioritized targets |
| --- | --- | --- |
| 4. Top-20 therapeutic shortlist | script · agent run | shortlist |
| --- | --- | --- |