From a Cell in a Well to an Indexed Library
Plate-based assays keep cells physically separated in wells during early processing. A cell is deposited into a designated well, its RNA is released, and the resulting cDNA is prepared as a library that can later be associated with that well. The central data-generation idea is physical separation followed by well-specific identification: the location of the cell helps preserve its identity when libraries are pooled for sequencing.
The earlier material on From Cell Capture to Gene Counts introduced the broad path from captured cells to gene-level measurements. Here the focus is the plate-based route through that path: what happens in a well, how a well becomes an identity label, and what “full-length” library preparation aims to capture. Unlike droplet workflows, where cell identity is associated with a droplet barcode, plate workflows can retain cell identity through the well assignment itself.
What Happens Inside the Well
Cells may be distributed into wells by fluorescence-activated cell sorting, manual transfer, or another controlled dispensing method. In sorting, an instrument can place a selected event into a specified well. The selection settings and the physical deposition are not proof that exactly one intact cell arrived, however. A well may be empty, may receive more than one cell, or may receive non-cellular material. The analysis therefore relies on the practical assumption that a well treated as a single-cell sample contains one cell and that its RNA is not materially mixed with RNA from another well.
Once a cell is in a well, it is lysed so that RNA is accessible for reverse transcription. Many full-length-oriented protocols use oligo-dT primers to initiate reverse transcription from the poly(A) tails of messenger RNAs. Some protocols use a template-switching step to add a common sequence at the other end of the cDNA. The resulting cDNA can then be amplified to provide enough material for library construction and sequencing. Exact reagents and reaction details vary across protocols; the shared principle is that RNA from a given well is processed separately long enough to preserve the connection between that material and the well.
Library preparation converts amplified cDNA into fragments and adds sequences needed for sequencing. A well-specific index is incorporated so that libraries from different wells can be pooled while retaining their origins. Depending on the protocol, this identity may be encoded through indexes added during library preparation, through well-specific processing, or through a combination of design choices. The essential requirement is that the index-to-well mapping be recorded correctly and remain distinguishable after pooling.
“Full-length” describes the intended coverage strategy, not a claim that every RNA molecule is copied from end to end. Reverse transcription can stop before reaching the other end of a transcript, amplification can favor some cDNAs over others, and library preparation can lose material. A full-length-oriented library may provide information across more of a transcript than a strongly end-focused assay, but coverage is still incomplete and uneven in practice.
How Well Identity Becomes Cell Identity
A plate layout is part of the data, not merely a laboratory convenience. Each well needs a unique, recorded link to the library made from its contents. Suppose well B03 is assigned to one cell and receives a library index reserved for B03. After libraries are pooled and sequenced, the index can be used to sort the sequencing reads back into the B03 library. If the cell was correctly deposited and the mapping was not confused, those reads are associated with the cell from B03.
This chain of identity depends on several conditions: the well assignment must be recorded accurately; the index must be sufficiently distinct from the indexes used for other wells; and library preparation or pooling must not cause substantial cross-contamination or index misassignment. A well label does not itself establish that the well contained exactly one cell. It identifies the source assigned to that library, subject to the validity of the physical and molecular steps.
The result is first a set of sequencing records organized by cell-associated library. Downstream processing can summarize those records as gene-level expression measurements for each cell. The measurement is not automatically a direct count of original RNA molecules: amplification and sequencing intervene between the RNA in the cell and the observed reads. The precise interpretation of read-derived measurements depends on the assay and later processing choices.
Worked Example: Tracking Wells Through a Small Plate
A hypothetical experiment deposits cells into 12 wells. After library preparation, the team records 10 wells with a library that passes its prespecified laboratory check. Of those 10 libraries, 9 are successfully assigned a distinct, documented index and are included in the sequencing pool. What fraction of the 12 wells contributes an indexed library to the pool?
The number of contributing wells is 9 and the starting number of wells is 12. The observed fraction is
Thus, 75% of the wells in this hypothetical plate contribute an indexed library to the pool. This is a descriptive yield for these wells, not an estimate that each well has a 75% chance of success in every experiment. The 2 wells without a passing library and the 1 passing library without a documented distinct index cannot be treated as successfully represented cell libraries in this pool.
Even for the 9 included libraries, the fraction does not establish that each library came from exactly one cell. That conclusion also depends on cell deposition and other quality checks.
What “Full-Length” Does and Does Not Tell You
A full-length-oriented approach is useful when the experimental goal calls for broader transcript coverage than a method focused on one end of a transcript. In broad terms, reverse transcription is designed to copy messenger RNA into cDNA in a way that can retain information from across transcript bodies, and amplification makes that cDNA available for library preparation. This design can support measurements that depend on sequence information outside a short terminal region.
The phrase should not be read as a guarantee that the entire original transcript is present in every library. For a transcript of length \(L\), it is possible for the protocol to generate cDNA covering only part of those \(L\) bases. Even when fragments from across a transcript are present in the library, sequencing samples only a portion of the prepared material. Therefore, a library described as full-length-oriented can still have gaps, unequal coverage, and no observed reads for transcripts that were present in the original cell.
Full-length protocols may also differ in how they convert reads into gene-level measurements. Some plate-based assays do not attach a unique label to each original RNA molecule. In those assays, amplification can make the number of observed reads differ from the number of original molecules. The presence of a well-specific index answers “which well did this library come from?” It does not, by itself, answer “how many original RNA molecules were present?”
Worked Example: Interpreting a Full-Length-Oriented Library
A hypothetical cell library contains reads aligned to several separated regions across a transcript that is 2,000 bases long. The reads cover a combined span of 1,350 bases, with gaps between some covered regions. What can be concluded from this observation?
The covered fraction of the transcript span is
The reads cover 67.5% of the transcript’s span in this example, with incomplete coverage. This is consistent with a full-length-oriented library that did not recover the entire transcript. It does not show that 67.5% of the original RNA molecules were captured, because the calculation describes covered sequence span, not the fraction of molecules recovered. Nor does it prove that all parts of the transcript were equally likely to be represented.
Checking the Identity Chain and Recognizing Failure
Plate-based workflows make some errors visible at the level of wells and libraries. A practical review follows the identity chain: which wells were intended to receive cells, which wells produced usable material, which indexes were assigned, and which libraries entered the pool. The following signs deserve attention:
- Many wells with no detectable library. This can result from an empty well, failed lysis or reverse transcription, loss of material, or a library-preparation failure. A missing library should not be interpreted as a cell with zero expression without supporting evidence.
- Unexpectedly high signal or mixed cell profiles in a well. These observations may be consistent with more than one cell in the well or contamination, although other technical and biological explanations are possible. A well label alone cannot rule out these possibilities.
- Indexes that are duplicated, missing, or inconsistently recorded. If two libraries cannot be distinguished reliably, reads may not be attributable to the intended well. The plate map and index records should be checked before pooling or interpreting cell-level results.
- Strongly uneven cDNA or transcript coverage. This can indicate uneven reverse transcription, amplification bias, or limited sequencing depth. “Full-length” should not be used to dismiss evident coverage gaps.
These checks are diagnostic clues, not automatic proof of a particular cause. For example, a low-yield well could have started with a small amount of RNA or could have experienced a processing failure. Distinguishing explanations requires protocol-specific controls and supporting observations.
How Plate Processing Differs in Organization
Both plate-based and droplet-based workflows aim to associate expression information with individual cells, but they preserve that association differently during capture. In a plate workflow, cells are assigned to spatially distinct wells, and well-specific processing or indexing connects a library to its source. In a droplet workflow, cell identity is associated with a droplet and its barcode. This is a contrast in organization and identity tracking, not a recommendation about which method to use; protocol choice depends on experimental aims and practical constraints.
Plate separation makes the well a natural unit for tracking processing history. It also means that an experiment’s records should preserve the plate layout and well-to-index mapping. By contrast, one should not infer from the phrase “plate-based” that all later measurement steps are identical across protocols. Library construction, amplification, sequencing, and quantification can vary. The later material on choosing between droplet and plate protocols addresses protocol selection; here the key point is how the plate workflow maintains cell-associated identity.
Common Mistakes and a Short Justification
A common mistake is to treat a well index as proof of a single cell, or to treat “full-length” as proof of complete transcript recovery. The correction follows from the actual chain of operations: an index labels a library’s assigned source, whereas one-cell occupancy depends on deposition; full-length design aims for broad transcript coverage, whereas actual recovery depends on reverse transcription, amplification, library construction, and sequencing.
The identity logic can be stated precisely. Let \(W\) denote the well assigned to a cell, and let \(I(W)\) denote the index recorded for the library prepared from that well. If each included well has a distinct index and the index mapping is recorded correctly, then observing index \(I(W)\) identifies reads as belonging to the library assigned to \(W\). This conclusion is about the library’s assigned origin. To infer that the reads represent one cell, an additional condition is needed: the well must have received one cell and the library must not have been materially contaminated.
The implication does not extend to cell occupancy without that additional condition. Keeping these claims separate prevents a bookkeeping label from being mistaken for a biological quality guarantee.
Worked Example: Interpreting a Pooled Plate Library
In a hypothetical plate, three wells are assigned indexes A, B, and C. Their libraries are pooled and sequenced. The resulting records include 4,800 reads with index A, 3,100 with index B, and 0 with index C. Which well-associated libraries have observed reads, and what cannot be concluded?
The index-to-well record links A to the first well and B to the second. Thus, the observed reads with indexes A and B are attributed to those two libraries, assuming correct indexing and no material index misassignment. No reads are observed for the library assigned index C in this sequencing result.
The read totals are \(4{,}800+3{,}100=7{,}900\) observed reads across A and B, and the observed read fraction for A among those two indexed groups is
So about 60.76% of these observed reads carry index A. This is a proportion of reads in this example, not a proportion of original RNA molecules. The absence of reads for C does not prove that its well was empty: its library may have failed, may have been too scarce to yield reads, or may have been excluded. The sequencing records alone do not distinguish these explanations.
Check Your Understanding
Answer each question, then compare your reasoning with the solutions.
- What does a well-specific index identify, and what additional condition is needed to interpret the library as coming from one cell?
- A plate has 48 wells, and 36 produce libraries that pass a laboratory check. What is the observed passing fraction?
- In a full-length-oriented assay, reads cover 900 bases of a 1,500-base transcript span. What fraction of the span is covered, and what does that fraction not measure?
- Why does observing 5,000 reads from one well not establish that the well contained 5,000 original RNA molecules?
- State one possible warning sign of a plate-based processing problem and one reason that sign does not prove a single cause.
Solutions. (1) The index identifies the library assigned to a well; interpreting it as one cell also requires that the well received one cell and was not materially contaminated. (2) \(36/48=0.75\), or 75%. (3) \(900/1{,}500=0.60\), or 60% of the span; this is not the fraction of original molecules recovered. (4) Amplification and sequencing intervene, so reads are not automatically one-to-one with original RNA molecules. (5) Many wells with no detectable library is one warning sign; it could arise from an empty well, failed processing, or loss of material, so it does not by itself identify the cause.