---
slideOptions:
transition: fade
theme: white
---
# Group Meeting 2019/09

###### tags: `Group Meeting`
<style>
code { border:1px outset #bbb; }
.reveal {
font-size: 24px;
}
</style>

---
# 2019/01/15

half-life: 15min~2h
Collect 30 samples' turnaround time from inex:
* Blood->Plasma
* Plasma->DNA

### Pearson

Pearson(FF vs. Blood->Plasma): -0.07
Pearson(FF vs. Plasma->DNA): -0.19


[notebook](http://xcnc45.comp.nus.edu.sg:8888/notebooks/INEX_564_predict.ipynb)

---
# 2019/12/13
**CAST**
#### Problem
2 sites far to each other, easy to get P>0.0001
#### Solution
1. convert `distance` to `recombination rate`
2. remove progeny with largest difference sum
3. do permutation test

**NIPT**
$(x,y) = \frac{\Sigma_{x \le i \le y}\ offset[i]}{total\ read}$

$(x,y) = \frac{\Sigma_{x \le i \le y}\ offset[i]}{\Sigma_{x-10 \le i \le y}\ offset[i]}$

---

### 2019/11/01
$\forall S \subset \mathbb{G}$
$~~~$ Let $\alpha=P(\text{read in } S|\text{fetal read})$
$~~~$ Let $\beta=P(\text{read in } S|\text{maternal read})$

---

For any sample:
$~~~~ \frac{\text{read count in } S}{\text{total read}}=f\alpha+(1-f)\beta ~~~~~~\Rightarrow~~~~~ f=\frac{r-\beta}{\alpha-\beta}$

> $\alpha$ and $\beta$ can be trained by linear fitting.

> chrY-based method is a special case where $S=chrY$

> **OBJECT: maximize $|\alpha-\beta|$ by good $S$**

---

nucleosome profile => read count around nucleosome centre
Pearson=0.73

---

#### Better nucleosome track
FFT (window size=600)

---


---

### 2019/09/20
Orientation-aware plasma cell-free DNA fragmentation analysis in open chromatin regions informs tissue of origin, ***Genome Research***, Dennis Lo.

Theory
Nucleosome position
Open Chromatin Region

### 2019/09/13
Sanefalcon vs Y-chrom
Sanefalcon profile
nucleosome size vs. first/second peak