Unit of cluster_count_pf?

[yfu-recursion]

Do you know where I can find the unit of cluster_count_pf. The value seems to suggest it is in thousands, but I can't seem to find the corresponding document...

I see 3 versions defined here:

interop/interop/model/metrics/tile_metric.h

Lines 334 to 349 in c98d268

	float cluster_count() const
	{ return m_cluster_count; }
	/** Number of kilo-clusters for each tile
	*
	* @return number of kilo-clusters
	*/
	float cluster_count_k() const
	{ return m_cluster_count/1000.0f; }
	/** Number of mega-clusters for each tile
	*
	* @return number of mega-clusters
	*/
	float cluster_count_m() const
	{ return m_cluster_count/1000000.0f; }

but it looks like run_summary() only has cluster_count_pf() available.

>>> run_metrics = py_interop_run_metrics.run_metrics()
>>> run_metrics.read(run_folder)
>>> summary = py_interop_summary.run_summary()
>>> summary.at(0).at(0).cluster_count_pf().median()
2682877.5

[ezralanglois]

What version of InterOp is this? Also, why instrument and consumable type?

This should be a count of clusters. It is strange that we get a floating point value, but that is due to the fact that the median is interpolated. Given that we get a fractional value, suggests this is a random platform like MiSeq or HiSeq.

[yfu-recursion]

Hi @ezralanglois! This is NovaSeq 6000 and S4. I am using interop 1.1.23.

$ pip list | grep interop
interop                     1.1.23

[ezralanglois]

Apologies, I confused myself with cluster count and PF cluster count.

For the run and real level, we report the sum of all the tiles, so you have cluster_count or cluster_count_pf

For the lane level, we report statistics over all the tiles. So median means the median cluster count PF for all tiles in that lane. The unit is clusters, but the value is the median number of clusters per tile

[yfu-recursion]

@ezralanglois Thank you for the detailed reply!